From 434b51e5c2fce756906dec4803900397bc98ad72 Mon Sep 17 00:00:00 2001 From: "Dr. David Alan Gilbert" Date: Mon, 27 Jan 2020 19:00:39 +0100 Subject: [PATCH 008/116] virtiofsd: Pull in upstream headers MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit RH-Author: Dr. David Alan Gilbert Message-id: <20200127190227.40942-5-dgilbert@redhat.com> Patchwork-id: 93457 O-Subject: [RHEL-AV-8.2 qemu-kvm PATCH 004/112] virtiofsd: Pull in upstream headers Bugzilla: 1694164 RH-Acked-by: Philippe Mathieu-Daudé RH-Acked-by: Stefan Hajnoczi RH-Acked-by: Sergio Lopez Pascual From: "Dr. David Alan Gilbert" Pull in headers fromlibfuse's upstream fuse-3.8.0 Signed-off-by: Dr. David Alan Gilbert Reviewed-by: Daniel P. Berrangé Reviewed-by: Philippe Mathieu-Daudé Signed-off-by: Dr. David Alan Gilbert (cherry picked from commit ee46c78901eb7fa78e328e04c0494ad6d207238b) Signed-off-by: Miroslav Rezanina --- tools/virtiofsd/fuse.h | 1275 ++++++++++++++++++++ tools/virtiofsd/fuse_common.h | 823 +++++++++++++ tools/virtiofsd/fuse_i.h | 139 +++ tools/virtiofsd/fuse_log.h | 82 ++ tools/virtiofsd/fuse_lowlevel.h | 2089 +++++++++++++++++++++++++++++++++ tools/virtiofsd/fuse_misc.h | 59 + tools/virtiofsd/fuse_opt.h | 271 +++++ tools/virtiofsd/passthrough_helpers.h | 76 ++ 8 files changed, 4814 insertions(+) create mode 100644 tools/virtiofsd/fuse.h create mode 100644 tools/virtiofsd/fuse_common.h create mode 100644 tools/virtiofsd/fuse_i.h create mode 100644 tools/virtiofsd/fuse_log.h create mode 100644 tools/virtiofsd/fuse_lowlevel.h create mode 100644 tools/virtiofsd/fuse_misc.h create mode 100644 tools/virtiofsd/fuse_opt.h create mode 100644 tools/virtiofsd/passthrough_helpers.h diff --git a/tools/virtiofsd/fuse.h b/tools/virtiofsd/fuse.h new file mode 100644 index 0000000..883f6e5 --- /dev/null +++ b/tools/virtiofsd/fuse.h @@ -0,0 +1,1275 @@ +/* + FUSE: Filesystem in Userspace + Copyright (C) 2001-2007 Miklos Szeredi + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB. +*/ + +#ifndef FUSE_H_ +#define FUSE_H_ + +/** @file + * + * This file defines the library interface of FUSE + * + * IMPORTANT: you should define FUSE_USE_VERSION before including this header. + */ + +#include "fuse_common.h" + +#include +#include +#include +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/* ----------------------------------------------------------- * + * Basic FUSE API * + * ----------------------------------------------------------- */ + +/** Handle for a FUSE filesystem */ +struct fuse; + +/** + * Readdir flags, passed to ->readdir() + */ +enum fuse_readdir_flags { + /** + * "Plus" mode. + * + * The kernel wants to prefill the inode cache during readdir. The + * filesystem may honour this by filling in the attributes and setting + * FUSE_FILL_DIR_FLAGS for the filler function. The filesystem may also + * just ignore this flag completely. + */ + FUSE_READDIR_PLUS = (1 << 0), +}; + +enum fuse_fill_dir_flags { + /** + * "Plus" mode: all file attributes are valid + * + * The attributes are used by the kernel to prefill the inode cache + * during a readdir. + * + * It is okay to set FUSE_FILL_DIR_PLUS if FUSE_READDIR_PLUS is not set + * and vice versa. + */ + FUSE_FILL_DIR_PLUS = (1 << 1), +}; + +/** Function to add an entry in a readdir() operation + * + * The *off* parameter can be any non-zero value that enables the + * filesystem to identify the current point in the directory + * stream. It does not need to be the actual physical position. A + * value of zero is reserved to indicate that seeking in directories + * is not supported. + * + * @param buf the buffer passed to the readdir() operation + * @param name the file name of the directory entry + * @param stat file attributes, can be NULL + * @param off offset of the next entry or zero + * @param flags fill flags + * @return 1 if buffer is full, zero otherwise + */ +typedef int (*fuse_fill_dir_t) (void *buf, const char *name, + const struct stat *stbuf, off_t off, + enum fuse_fill_dir_flags flags); +/** + * Configuration of the high-level API + * + * This structure is initialized from the arguments passed to + * fuse_new(), and then passed to the file system's init() handler + * which should ensure that the configuration is compatible with the + * file system implementation. + */ +struct fuse_config { + /** + * If `set_gid` is non-zero, the st_gid attribute of each file + * is overwritten with the value of `gid`. + */ + int set_gid; + unsigned int gid; + + /** + * If `set_uid` is non-zero, the st_uid attribute of each file + * is overwritten with the value of `uid`. + */ + int set_uid; + unsigned int uid; + + /** + * If `set_mode` is non-zero, the any permissions bits set in + * `umask` are unset in the st_mode attribute of each file. + */ + int set_mode; + unsigned int umask; + + /** + * The timeout in seconds for which name lookups will be + * cached. + */ + double entry_timeout; + + /** + * The timeout in seconds for which a negative lookup will be + * cached. This means, that if file did not exist (lookup + * retuned ENOENT), the lookup will only be redone after the + * timeout, and the file/directory will be assumed to not + * exist until then. A value of zero means that negative + * lookups are not cached. + */ + double negative_timeout; + + /** + * The timeout in seconds for which file/directory attributes + * (as returned by e.g. the `getattr` handler) are cached. + */ + double attr_timeout; + + /** + * Allow requests to be interrupted + */ + int intr; + + /** + * Specify which signal number to send to the filesystem when + * a request is interrupted. The default is hardcoded to + * USR1. + */ + int intr_signal; + + /** + * Normally, FUSE assigns inodes to paths only for as long as + * the kernel is aware of them. With this option inodes are + * instead remembered for at least this many seconds. This + * will require more memory, but may be necessary when using + * applications that make use of inode numbers. + * + * A number of -1 means that inodes will be remembered for the + * entire life-time of the file-system process. + */ + int remember; + + /** + * The default behavior is that if an open file is deleted, + * the file is renamed to a hidden file (.fuse_hiddenXXX), and + * only removed when the file is finally released. This + * relieves the filesystem implementation of having to deal + * with this problem. This option disables the hiding + * behavior, and files are removed immediately in an unlink + * operation (or in a rename operation which overwrites an + * existing file). + * + * It is recommended that you not use the hard_remove + * option. When hard_remove is set, the following libc + * functions fail on unlinked files (returning errno of + * ENOENT): read(2), write(2), fsync(2), close(2), f*xattr(2), + * ftruncate(2), fstat(2), fchmod(2), fchown(2) + */ + int hard_remove; + + /** + * Honor the st_ino field in the functions getattr() and + * fill_dir(). This value is used to fill in the st_ino field + * in the stat(2), lstat(2), fstat(2) functions and the d_ino + * field in the readdir(2) function. The filesystem does not + * have to guarantee uniqueness, however some applications + * rely on this value being unique for the whole filesystem. + * + * Note that this does *not* affect the inode that libfuse + * and the kernel use internally (also called the "nodeid"). + */ + int use_ino; + + /** + * If use_ino option is not given, still try to fill in the + * d_ino field in readdir(2). If the name was previously + * looked up, and is still in the cache, the inode number + * found there will be used. Otherwise it will be set to -1. + * If use_ino option is given, this option is ignored. + */ + int readdir_ino; + + /** + * This option disables the use of page cache (file content cache) + * in the kernel for this filesystem. This has several affects: + * + * 1. Each read(2) or write(2) system call will initiate one + * or more read or write operations, data will not be + * cached in the kernel. + * + * 2. The return value of the read() and write() system calls + * will correspond to the return values of the read and + * write operations. This is useful for example if the + * file size is not known in advance (before reading it). + * + * Internally, enabling this option causes fuse to set the + * `direct_io` field of `struct fuse_file_info` - overwriting + * any value that was put there by the file system. + */ + int direct_io; + + /** + * This option disables flushing the cache of the file + * contents on every open(2). This should only be enabled on + * filesystems where the file data is never changed + * externally (not through the mounted FUSE filesystem). Thus + * it is not suitable for network filesystems and other + * intermediate filesystems. + * + * NOTE: if this option is not specified (and neither + * direct_io) data is still cached after the open(2), so a + * read(2) system call will not always initiate a read + * operation. + * + * Internally, enabling this option causes fuse to set the + * `keep_cache` field of `struct fuse_file_info` - overwriting + * any value that was put there by the file system. + */ + int kernel_cache; + + /** + * This option is an alternative to `kernel_cache`. Instead of + * unconditionally keeping cached data, the cached data is + * invalidated on open(2) if if the modification time or the + * size of the file has changed since it was last opened. + */ + int auto_cache; + + /** + * The timeout in seconds for which file attributes are cached + * for the purpose of checking if auto_cache should flush the + * file data on open. + */ + int ac_attr_timeout_set; + double ac_attr_timeout; + + /** + * If this option is given the file-system handlers for the + * following operations will not receive path information: + * read, write, flush, release, fsync, readdir, releasedir, + * fsyncdir, lock, ioctl and poll. + * + * For the truncate, getattr, chmod, chown and utimens + * operations the path will be provided only if the struct + * fuse_file_info argument is NULL. + */ + int nullpath_ok; + + /** + * The remaining options are used by libfuse internally and + * should not be touched. + */ + int show_help; + char *modules; + int debug; +}; + + +/** + * The file system operations: + * + * Most of these should work very similarly to the well known UNIX + * file system operations. A major exception is that instead of + * returning an error in 'errno', the operation should return the + * negated error value (-errno) directly. + * + * All methods are optional, but some are essential for a useful + * filesystem (e.g. getattr). Open, flush, release, fsync, opendir, + * releasedir, fsyncdir, access, create, truncate, lock, init and + * destroy are special purpose methods, without which a full featured + * filesystem can still be implemented. + * + * In general, all methods are expected to perform any necessary + * permission checking. However, a filesystem may delegate this task + * to the kernel by passing the `default_permissions` mount option to + * `fuse_new()`. In this case, methods will only be called if + * the kernel's permission check has succeeded. + * + * Almost all operations take a path which can be of any length. + */ +struct fuse_operations { + /** Get file attributes. + * + * Similar to stat(). The 'st_dev' and 'st_blksize' fields are + * ignored. The 'st_ino' field is ignored except if the 'use_ino' + * mount option is given. In that case it is passed to userspace, + * but libfuse and the kernel will still assign a different + * inode for internal use (called the "nodeid"). + * + * `fi` will always be NULL if the file is not currently open, but + * may also be NULL if the file is open. + */ + int (*getattr) (const char *, struct stat *, struct fuse_file_info *fi); + + /** Read the target of a symbolic link + * + * The buffer should be filled with a null terminated string. The + * buffer size argument includes the space for the terminating + * null character. If the linkname is too long to fit in the + * buffer, it should be truncated. The return value should be 0 + * for success. + */ + int (*readlink) (const char *, char *, size_t); + + /** Create a file node + * + * This is called for creation of all non-directory, non-symlink + * nodes. If the filesystem defines a create() method, then for + * regular files that will be called instead. + */ + int (*mknod) (const char *, mode_t, dev_t); + + /** Create a directory + * + * Note that the mode argument may not have the type specification + * bits set, i.e. S_ISDIR(mode) can be false. To obtain the + * correct directory type bits use mode|S_IFDIR + * */ + int (*mkdir) (const char *, mode_t); + + /** Remove a file */ + int (*unlink) (const char *); + + /** Remove a directory */ + int (*rmdir) (const char *); + + /** Create a symbolic link */ + int (*symlink) (const char *, const char *); + + /** Rename a file + * + * *flags* may be `RENAME_EXCHANGE` or `RENAME_NOREPLACE`. If + * RENAME_NOREPLACE is specified, the filesystem must not + * overwrite *newname* if it exists and return an error + * instead. If `RENAME_EXCHANGE` is specified, the filesystem + * must atomically exchange the two files, i.e. both must + * exist and neither may be deleted. + */ + int (*rename) (const char *, const char *, unsigned int flags); + + /** Create a hard link to a file */ + int (*link) (const char *, const char *); + + /** Change the permission bits of a file + * + * `fi` will always be NULL if the file is not currenlty open, but + * may also be NULL if the file is open. + */ + int (*chmod) (const char *, mode_t, struct fuse_file_info *fi); + + /** Change the owner and group of a file + * + * `fi` will always be NULL if the file is not currenlty open, but + * may also be NULL if the file is open. + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits. + */ + int (*chown) (const char *, uid_t, gid_t, struct fuse_file_info *fi); + + /** Change the size of a file + * + * `fi` will always be NULL if the file is not currenlty open, but + * may also be NULL if the file is open. + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits. + */ + int (*truncate) (const char *, off_t, struct fuse_file_info *fi); + + /** Open a file + * + * Open flags are available in fi->flags. The following rules + * apply. + * + * - Creation (O_CREAT, O_EXCL, O_NOCTTY) flags will be + * filtered out / handled by the kernel. + * + * - Access modes (O_RDONLY, O_WRONLY, O_RDWR, O_EXEC, O_SEARCH) + * should be used by the filesystem to check if the operation is + * permitted. If the ``-o default_permissions`` mount option is + * given, this check is already done by the kernel before calling + * open() and may thus be omitted by the filesystem. + * + * - When writeback caching is enabled, the kernel may send + * read requests even for files opened with O_WRONLY. The + * filesystem should be prepared to handle this. + * + * - When writeback caching is disabled, the filesystem is + * expected to properly handle the O_APPEND flag and ensure + * that each write is appending to the end of the file. + * + * - When writeback caching is enabled, the kernel will + * handle O_APPEND. However, unless all changes to the file + * come through the kernel this will not work reliably. The + * filesystem should thus either ignore the O_APPEND flag + * (and let the kernel handle it), or return an error + * (indicating that reliably O_APPEND is not available). + * + * Filesystem may store an arbitrary file handle (pointer, + * index, etc) in fi->fh, and use this in other all other file + * operations (read, write, flush, release, fsync). + * + * Filesystem may also implement stateless file I/O and not store + * anything in fi->fh. + * + * There are also some flags (direct_io, keep_cache) which the + * filesystem may set in fi, to change the way the file is opened. + * See fuse_file_info structure in for more details. + * + * If this request is answered with an error code of ENOSYS + * and FUSE_CAP_NO_OPEN_SUPPORT is set in + * `fuse_conn_info.capable`, this is treated as success and + * future calls to open will also succeed without being send + * to the filesystem process. + * + */ + int (*open) (const char *, struct fuse_file_info *); + + /** Read data from an open file + * + * Read should return exactly the number of bytes requested except + * on EOF or error, otherwise the rest of the data will be + * substituted with zeroes. An exception to this is when the + * 'direct_io' mount option is specified, in which case the return + * value of the read system call will reflect the return value of + * this operation. + */ + int (*read) (const char *, char *, size_t, off_t, + struct fuse_file_info *); + + /** Write data to an open file + * + * Write should return exactly the number of bytes requested + * except on error. An exception to this is when the 'direct_io' + * mount option is specified (see read operation). + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits. + */ + int (*write) (const char *, const char *, size_t, off_t, + struct fuse_file_info *); + + /** Get file system statistics + * + * The 'f_favail', 'f_fsid' and 'f_flag' fields are ignored + */ + int (*statfs) (const char *, struct statvfs *); + + /** Possibly flush cached data + * + * BIG NOTE: This is not equivalent to fsync(). It's not a + * request to sync dirty data. + * + * Flush is called on each close() of a file descriptor, as opposed to + * release which is called on the close of the last file descriptor for + * a file. Under Linux, errors returned by flush() will be passed to + * userspace as errors from close(), so flush() is a good place to write + * back any cached dirty data. However, many applications ignore errors + * on close(), and on non-Linux systems, close() may succeed even if flush() + * returns an error. For these reasons, filesystems should not assume + * that errors returned by flush will ever be noticed or even + * delivered. + * + * NOTE: The flush() method may be called more than once for each + * open(). This happens if more than one file descriptor refers to an + * open file handle, e.g. due to dup(), dup2() or fork() calls. It is + * not possible to determine if a flush is final, so each flush should + * be treated equally. Multiple write-flush sequences are relatively + * rare, so this shouldn't be a problem. + * + * Filesystems shouldn't assume that flush will be called at any + * particular point. It may be called more times than expected, or not + * at all. + * + * [close]: http://pubs.opengroup.org/onlinepubs/9699919799/functions/close.html + */ + int (*flush) (const char *, struct fuse_file_info *); + + /** Release an open file + * + * Release is called when there are no more references to an open + * file: all file descriptors are closed and all memory mappings + * are unmapped. + * + * For every open() call there will be exactly one release() call + * with the same flags and file handle. It is possible to + * have a file opened more than once, in which case only the last + * release will mean, that no more reads/writes will happen on the + * file. The return value of release is ignored. + */ + int (*release) (const char *, struct fuse_file_info *); + + /** Synchronize file contents + * + * If the datasync parameter is non-zero, then only the user data + * should be flushed, not the meta data. + */ + int (*fsync) (const char *, int, struct fuse_file_info *); + + /** Set extended attributes */ + int (*setxattr) (const char *, const char *, const char *, size_t, int); + + /** Get extended attributes */ + int (*getxattr) (const char *, const char *, char *, size_t); + + /** List extended attributes */ + int (*listxattr) (const char *, char *, size_t); + + /** Remove extended attributes */ + int (*removexattr) (const char *, const char *); + + /** Open directory + * + * Unless the 'default_permissions' mount option is given, + * this method should check if opendir is permitted for this + * directory. Optionally opendir may also return an arbitrary + * filehandle in the fuse_file_info structure, which will be + * passed to readdir, releasedir and fsyncdir. + */ + int (*opendir) (const char *, struct fuse_file_info *); + + /** Read directory + * + * The filesystem may choose between two modes of operation: + * + * 1) The readdir implementation ignores the offset parameter, and + * passes zero to the filler function's offset. The filler + * function will not return '1' (unless an error happens), so the + * whole directory is read in a single readdir operation. + * + * 2) The readdir implementation keeps track of the offsets of the + * directory entries. It uses the offset parameter and always + * passes non-zero offset to the filler function. When the buffer + * is full (or an error happens) the filler function will return + * '1'. + */ + int (*readdir) (const char *, void *, fuse_fill_dir_t, off_t, + struct fuse_file_info *, enum fuse_readdir_flags); + + /** Release directory + */ + int (*releasedir) (const char *, struct fuse_file_info *); + + /** Synchronize directory contents + * + * If the datasync parameter is non-zero, then only the user data + * should be flushed, not the meta data + */ + int (*fsyncdir) (const char *, int, struct fuse_file_info *); + + /** + * Initialize filesystem + * + * The return value will passed in the `private_data` field of + * `struct fuse_context` to all file operations, and as a + * parameter to the destroy() method. It overrides the initial + * value provided to fuse_main() / fuse_new(). + */ + void *(*init) (struct fuse_conn_info *conn, + struct fuse_config *cfg); + + /** + * Clean up filesystem + * + * Called on filesystem exit. + */ + void (*destroy) (void *private_data); + + /** + * Check file access permissions + * + * This will be called for the access() system call. If the + * 'default_permissions' mount option is given, this method is not + * called. + * + * This method is not called under Linux kernel versions 2.4.x + */ + int (*access) (const char *, int); + + /** + * Create and open a file + * + * If the file does not exist, first create it with the specified + * mode, and then open it. + * + * If this method is not implemented or under Linux kernel + * versions earlier than 2.6.15, the mknod() and open() methods + * will be called instead. + */ + int (*create) (const char *, mode_t, struct fuse_file_info *); + + /** + * Perform POSIX file locking operation + * + * The cmd argument will be either F_GETLK, F_SETLK or F_SETLKW. + * + * For the meaning of fields in 'struct flock' see the man page + * for fcntl(2). The l_whence field will always be set to + * SEEK_SET. + * + * For checking lock ownership, the 'fuse_file_info->owner' + * argument must be used. + * + * For F_GETLK operation, the library will first check currently + * held locks, and if a conflicting lock is found it will return + * information without calling this method. This ensures, that + * for local locks the l_pid field is correctly filled in. The + * results may not be accurate in case of race conditions and in + * the presence of hard links, but it's unlikely that an + * application would rely on accurate GETLK results in these + * cases. If a conflicting lock is not found, this method will be + * called, and the filesystem may fill out l_pid by a meaningful + * value, or it may leave this field zero. + * + * For F_SETLK and F_SETLKW the l_pid field will be set to the pid + * of the process performing the locking operation. + * + * Note: if this method is not implemented, the kernel will still + * allow file locking to work locally. Hence it is only + * interesting for network filesystems and similar. + */ + int (*lock) (const char *, struct fuse_file_info *, int cmd, + struct flock *); + + /** + * Change the access and modification times of a file with + * nanosecond resolution + * + * This supersedes the old utime() interface. New applications + * should use this. + * + * `fi` will always be NULL if the file is not currenlty open, but + * may also be NULL if the file is open. + * + * See the utimensat(2) man page for details. + */ + int (*utimens) (const char *, const struct timespec tv[2], + struct fuse_file_info *fi); + + /** + * Map block index within file to block index within device + * + * Note: This makes sense only for block device backed filesystems + * mounted with the 'blkdev' option + */ + int (*bmap) (const char *, size_t blocksize, uint64_t *idx); + + /** + * Ioctl + * + * flags will have FUSE_IOCTL_COMPAT set for 32bit ioctls in + * 64bit environment. The size and direction of data is + * determined by _IOC_*() decoding of cmd. For _IOC_NONE, + * data will be NULL, for _IOC_WRITE data is out area, for + * _IOC_READ in area and if both are set in/out area. In all + * non-NULL cases, the area is of _IOC_SIZE(cmd) bytes. + * + * If flags has FUSE_IOCTL_DIR then the fuse_file_info refers to a + * directory file handle. + * + * Note : the unsigned long request submitted by the application + * is truncated to 32 bits. + */ + int (*ioctl) (const char *, unsigned int cmd, void *arg, + struct fuse_file_info *, unsigned int flags, void *data); + + /** + * Poll for IO readiness events + * + * Note: If ph is non-NULL, the client should notify + * when IO readiness events occur by calling + * fuse_notify_poll() with the specified ph. + * + * Regardless of the number of times poll with a non-NULL ph + * is received, single notification is enough to clear all. + * Notifying more times incurs overhead but doesn't harm + * correctness. + * + * The callee is responsible for destroying ph with + * fuse_pollhandle_destroy() when no longer in use. + */ + int (*poll) (const char *, struct fuse_file_info *, + struct fuse_pollhandle *ph, unsigned *reventsp); + + /** Write contents of buffer to an open file + * + * Similar to the write() method, but data is supplied in a + * generic buffer. Use fuse_buf_copy() to transfer data to + * the destination. + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits. + */ + int (*write_buf) (const char *, struct fuse_bufvec *buf, off_t off, + struct fuse_file_info *); + + /** Store data from an open file in a buffer + * + * Similar to the read() method, but data is stored and + * returned in a generic buffer. + * + * No actual copying of data has to take place, the source + * file descriptor may simply be stored in the buffer for + * later data transfer. + * + * The buffer must be allocated dynamically and stored at the + * location pointed to by bufp. If the buffer contains memory + * regions, they too must be allocated using malloc(). The + * allocated memory will be freed by the caller. + */ + int (*read_buf) (const char *, struct fuse_bufvec **bufp, + size_t size, off_t off, struct fuse_file_info *); + /** + * Perform BSD file locking operation + * + * The op argument will be either LOCK_SH, LOCK_EX or LOCK_UN + * + * Nonblocking requests will be indicated by ORing LOCK_NB to + * the above operations + * + * For more information see the flock(2) manual page. + * + * Additionally fi->owner will be set to a value unique to + * this open file. This same value will be supplied to + * ->release() when the file is released. + * + * Note: if this method is not implemented, the kernel will still + * allow file locking to work locally. Hence it is only + * interesting for network filesystems and similar. + */ + int (*flock) (const char *, struct fuse_file_info *, int op); + + /** + * Allocates space for an open file + * + * This function ensures that required space is allocated for specified + * file. If this function returns success then any subsequent write + * request to specified range is guaranteed not to fail because of lack + * of space on the file system media. + */ + int (*fallocate) (const char *, int, off_t, off_t, + struct fuse_file_info *); + + /** + * Copy a range of data from one file to another + * + * Performs an optimized copy between two file descriptors without the + * additional cost of transferring data through the FUSE kernel module + * to user space (glibc) and then back into the FUSE filesystem again. + * + * In case this method is not implemented, glibc falls back to reading + * data from the source and writing to the destination. Effectively + * doing an inefficient copy of the data. + */ + ssize_t (*copy_file_range) (const char *path_in, + struct fuse_file_info *fi_in, + off_t offset_in, const char *path_out, + struct fuse_file_info *fi_out, + off_t offset_out, size_t size, int flags); + + /** + * Find next data or hole after the specified offset + */ + off_t (*lseek) (const char *, off_t off, int whence, struct fuse_file_info *); +}; + +/** Extra context that may be needed by some filesystems + * + * The uid, gid and pid fields are not filled in case of a writepage + * operation. + */ +struct fuse_context { + /** Pointer to the fuse object */ + struct fuse *fuse; + + /** User ID of the calling process */ + uid_t uid; + + /** Group ID of the calling process */ + gid_t gid; + + /** Process ID of the calling thread */ + pid_t pid; + + /** Private filesystem data */ + void *private_data; + + /** Umask of the calling process */ + mode_t umask; +}; + +/** + * Main function of FUSE. + * + * This is for the lazy. This is all that has to be called from the + * main() function. + * + * This function does the following: + * - parses command line options, and handles --help and + * --version + * - installs signal handlers for INT, HUP, TERM and PIPE + * - registers an exit handler to unmount the filesystem on program exit + * - creates a fuse handle + * - registers the operations + * - calls either the single-threaded or the multi-threaded event loop + * + * Most file systems will have to parse some file-system specific + * arguments before calling this function. It is recommended to do + * this with fuse_opt_parse() and a processing function that passes + * through any unknown options (this can also be achieved by just + * passing NULL as the processing function). That way, the remaining + * options can be passed directly to fuse_main(). + * + * fuse_main() accepts all options that can be passed to + * fuse_parse_cmdline(), fuse_new(), or fuse_session_new(). + * + * Option parsing skips argv[0], which is assumed to contain the + * program name. This element must always be present and is used to + * construct a basic ``usage: `` message for the --help + * output. argv[0] may also be set to the empty string. In this case + * the usage message is suppressed. This can be used by file systems + * to print their own usage line first. See hello.c for an example of + * how to do this. + * + * Note: this is currently implemented as a macro. + * + * The following error codes may be returned from fuse_main(): + * 1: Invalid option arguments + * 2: No mount point specified + * 3: FUSE setup failed + * 4: Mounting failed + * 5: Failed to daemonize (detach from session) + * 6: Failed to set up signal handlers + * 7: An error occured during the life of the file system + * + * @param argc the argument counter passed to the main() function + * @param argv the argument vector passed to the main() function + * @param op the file system operation + * @param private_data Initial value for the `private_data` + * field of `struct fuse_context`. May be overridden by the + * `struct fuse_operations.init` handler. + * @return 0 on success, nonzero on failure + * + * Example usage, see hello.c + */ +/* + int fuse_main(int argc, char *argv[], const struct fuse_operations *op, + void *private_data); +*/ +#define fuse_main(argc, argv, op, private_data) \ + fuse_main_real(argc, argv, op, sizeof(*(op)), private_data) + +/* ----------------------------------------------------------- * + * More detailed API * + * ----------------------------------------------------------- */ + +/** + * Print available options (high- and low-level) to stdout. This is + * not an exhaustive list, but includes only those options that may be + * of interest to an end-user of a file system. + * + * The function looks at the argument vector only to determine if + * there are additional modules to be loaded (module=foo option), + * and attempts to call their help functions as well. + * + * @param args the argument vector. + */ +void fuse_lib_help(struct fuse_args *args); + +/** + * Create a new FUSE filesystem. + * + * This function accepts most file-system independent mount options + * (like context, nodev, ro - see mount(8)), as well as the + * FUSE-specific mount options from mount.fuse(8). + * + * If the --help option is specified, the function writes a help text + * to stdout and returns NULL. + * + * Option parsing skips argv[0], which is assumed to contain the + * program name. This element must always be present and is used to + * construct a basic ``usage: `` message for the --help output. If + * argv[0] is set to the empty string, no usage message is included in + * the --help output. + * + * If an unknown option is passed in, an error message is written to + * stderr and the function returns NULL. + * + * @param args argument vector + * @param op the filesystem operations + * @param op_size the size of the fuse_operations structure + * @param private_data Initial value for the `private_data` + * field of `struct fuse_context`. May be overridden by the + * `struct fuse_operations.init` handler. + * @return the created FUSE handle + */ +#if FUSE_USE_VERSION == 30 +struct fuse *fuse_new_30(struct fuse_args *args, const struct fuse_operations *op, + size_t op_size, void *private_data); +#define fuse_new(args, op, size, data) fuse_new_30(args, op, size, data) +#else +struct fuse *fuse_new(struct fuse_args *args, const struct fuse_operations *op, + size_t op_size, void *private_data); +#endif + +/** + * Mount a FUSE file system. + * + * @param mountpoint the mount point path + * @param f the FUSE handle + * + * @return 0 on success, -1 on failure. + **/ +int fuse_mount(struct fuse *f, const char *mountpoint); + +/** + * Unmount a FUSE file system. + * + * See fuse_session_unmount() for additional information. + * + * @param f the FUSE handle + **/ +void fuse_unmount(struct fuse *f); + +/** + * Destroy the FUSE handle. + * + * NOTE: This function does not unmount the filesystem. If this is + * needed, call fuse_unmount() before calling this function. + * + * @param f the FUSE handle + */ +void fuse_destroy(struct fuse *f); + +/** + * FUSE event loop. + * + * Requests from the kernel are processed, and the appropriate + * operations are called. + * + * For a description of the return value and the conditions when the + * event loop exits, refer to the documentation of + * fuse_session_loop(). + * + * @param f the FUSE handle + * @return see fuse_session_loop() + * + * See also: fuse_loop_mt() + */ +int fuse_loop(struct fuse *f); + +/** + * Flag session as terminated + * + * This function will cause any running event loops to exit on + * the next opportunity. + * + * @param f the FUSE handle + */ +void fuse_exit(struct fuse *f); + +/** + * FUSE event loop with multiple threads + * + * Requests from the kernel are processed, and the appropriate + * operations are called. Request are processed in parallel by + * distributing them between multiple threads. + * + * For a description of the return value and the conditions when the + * event loop exits, refer to the documentation of + * fuse_session_loop(). + * + * Note: using fuse_loop() instead of fuse_loop_mt() means you are running in + * single-threaded mode, and that you will not have to worry about reentrancy, + * though you will have to worry about recursive lookups. In single-threaded + * mode, FUSE will wait for one callback to return before calling another. + * + * Enabling multiple threads, by using fuse_loop_mt(), will cause FUSE to make + * multiple simultaneous calls into the various callback functions given by your + * fuse_operations record. + * + * If you are using multiple threads, you can enjoy all the parallel execution + * and interactive response benefits of threads, and you get to enjoy all the + * benefits of race conditions and locking bugs, too. Ensure that any code used + * in the callback function of fuse_operations is also thread-safe. + * + * @param f the FUSE handle + * @param config loop configuration + * @return see fuse_session_loop() + * + * See also: fuse_loop() + */ +#if FUSE_USE_VERSION < 32 +int fuse_loop_mt_31(struct fuse *f, int clone_fd); +#define fuse_loop_mt(f, clone_fd) fuse_loop_mt_31(f, clone_fd) +#else +int fuse_loop_mt(struct fuse *f, struct fuse_loop_config *config); +#endif + +/** + * Get the current context + * + * The context is only valid for the duration of a filesystem + * operation, and thus must not be stored and used later. + * + * @return the context + */ +struct fuse_context *fuse_get_context(void); + +/** + * Get the current supplementary group IDs for the current request + * + * Similar to the getgroups(2) system call, except the return value is + * always the total number of group IDs, even if it is larger than the + * specified size. + * + * The current fuse kernel module in linux (as of 2.6.30) doesn't pass + * the group list to userspace, hence this function needs to parse + * "/proc/$TID/task/$TID/status" to get the group IDs. + * + * This feature may not be supported on all operating systems. In + * such a case this function will return -ENOSYS. + * + * @param size size of given array + * @param list array of group IDs to be filled in + * @return the total number of supplementary group IDs or -errno on failure + */ +int fuse_getgroups(int size, gid_t list[]); + +/** + * Check if the current request has already been interrupted + * + * @return 1 if the request has been interrupted, 0 otherwise + */ +int fuse_interrupted(void); + +/** + * Invalidates cache for the given path. + * + * This calls fuse_lowlevel_notify_inval_inode internally. + * + * @return 0 on successful invalidation, negative error value otherwise. + * This routine may return -ENOENT to indicate that there was + * no entry to be invalidated, e.g., because the path has not + * been seen before or has been forgotten; this should not be + * considered to be an error. + */ +int fuse_invalidate_path(struct fuse *f, const char *path); + +/** + * The real main function + * + * Do not call this directly, use fuse_main() + */ +int fuse_main_real(int argc, char *argv[], const struct fuse_operations *op, + size_t op_size, void *private_data); + +/** + * Start the cleanup thread when using option "remember". + * + * This is done automatically by fuse_loop_mt() + * @param fuse struct fuse pointer for fuse instance + * @return 0 on success and -1 on error + */ +int fuse_start_cleanup_thread(struct fuse *fuse); + +/** + * Stop the cleanup thread when using option "remember". + * + * This is done automatically by fuse_loop_mt() + * @param fuse struct fuse pointer for fuse instance + */ +void fuse_stop_cleanup_thread(struct fuse *fuse); + +/** + * Iterate over cache removing stale entries + * use in conjunction with "-oremember" + * + * NOTE: This is already done for the standard sessions + * + * @param fuse struct fuse pointer for fuse instance + * @return the number of seconds until the next cleanup + */ +int fuse_clean_cache(struct fuse *fuse); + +/* + * Stacking API + */ + +/** + * Fuse filesystem object + * + * This is opaque object represents a filesystem layer + */ +struct fuse_fs; + +/* + * These functions call the relevant filesystem operation, and return + * the result. + * + * If the operation is not defined, they return -ENOSYS, with the + * exception of fuse_fs_open, fuse_fs_release, fuse_fs_opendir, + * fuse_fs_releasedir and fuse_fs_statfs, which return 0. + */ + +int fuse_fs_getattr(struct fuse_fs *fs, const char *path, struct stat *buf, + struct fuse_file_info *fi); +int fuse_fs_rename(struct fuse_fs *fs, const char *oldpath, + const char *newpath, unsigned int flags); +int fuse_fs_unlink(struct fuse_fs *fs, const char *path); +int fuse_fs_rmdir(struct fuse_fs *fs, const char *path); +int fuse_fs_symlink(struct fuse_fs *fs, const char *linkname, + const char *path); +int fuse_fs_link(struct fuse_fs *fs, const char *oldpath, const char *newpath); +int fuse_fs_release(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi); +int fuse_fs_open(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi); +int fuse_fs_read(struct fuse_fs *fs, const char *path, char *buf, size_t size, + off_t off, struct fuse_file_info *fi); +int fuse_fs_read_buf(struct fuse_fs *fs, const char *path, + struct fuse_bufvec **bufp, size_t size, off_t off, + struct fuse_file_info *fi); +int fuse_fs_write(struct fuse_fs *fs, const char *path, const char *buf, + size_t size, off_t off, struct fuse_file_info *fi); +int fuse_fs_write_buf(struct fuse_fs *fs, const char *path, + struct fuse_bufvec *buf, off_t off, + struct fuse_file_info *fi); +int fuse_fs_fsync(struct fuse_fs *fs, const char *path, int datasync, + struct fuse_file_info *fi); +int fuse_fs_flush(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi); +int fuse_fs_statfs(struct fuse_fs *fs, const char *path, struct statvfs *buf); +int fuse_fs_opendir(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi); +int fuse_fs_readdir(struct fuse_fs *fs, const char *path, void *buf, + fuse_fill_dir_t filler, off_t off, + struct fuse_file_info *fi, enum fuse_readdir_flags flags); +int fuse_fs_fsyncdir(struct fuse_fs *fs, const char *path, int datasync, + struct fuse_file_info *fi); +int fuse_fs_releasedir(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi); +int fuse_fs_create(struct fuse_fs *fs, const char *path, mode_t mode, + struct fuse_file_info *fi); +int fuse_fs_lock(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi, int cmd, struct flock *lock); +int fuse_fs_flock(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi, int op); +int fuse_fs_chmod(struct fuse_fs *fs, const char *path, mode_t mode, + struct fuse_file_info *fi); +int fuse_fs_chown(struct fuse_fs *fs, const char *path, uid_t uid, gid_t gid, + struct fuse_file_info *fi); +int fuse_fs_truncate(struct fuse_fs *fs, const char *path, off_t size, + struct fuse_file_info *fi); +int fuse_fs_utimens(struct fuse_fs *fs, const char *path, + const struct timespec tv[2], struct fuse_file_info *fi); +int fuse_fs_access(struct fuse_fs *fs, const char *path, int mask); +int fuse_fs_readlink(struct fuse_fs *fs, const char *path, char *buf, + size_t len); +int fuse_fs_mknod(struct fuse_fs *fs, const char *path, mode_t mode, + dev_t rdev); +int fuse_fs_mkdir(struct fuse_fs *fs, const char *path, mode_t mode); +int fuse_fs_setxattr(struct fuse_fs *fs, const char *path, const char *name, + const char *value, size_t size, int flags); +int fuse_fs_getxattr(struct fuse_fs *fs, const char *path, const char *name, + char *value, size_t size); +int fuse_fs_listxattr(struct fuse_fs *fs, const char *path, char *list, + size_t size); +int fuse_fs_removexattr(struct fuse_fs *fs, const char *path, + const char *name); +int fuse_fs_bmap(struct fuse_fs *fs, const char *path, size_t blocksize, + uint64_t *idx); +int fuse_fs_ioctl(struct fuse_fs *fs, const char *path, unsigned int cmd, + void *arg, struct fuse_file_info *fi, unsigned int flags, + void *data); +int fuse_fs_poll(struct fuse_fs *fs, const char *path, + struct fuse_file_info *fi, struct fuse_pollhandle *ph, + unsigned *reventsp); +int fuse_fs_fallocate(struct fuse_fs *fs, const char *path, int mode, + off_t offset, off_t length, struct fuse_file_info *fi); +ssize_t fuse_fs_copy_file_range(struct fuse_fs *fs, const char *path_in, + struct fuse_file_info *fi_in, off_t off_in, + const char *path_out, + struct fuse_file_info *fi_out, off_t off_out, + size_t len, int flags); +off_t fuse_fs_lseek(struct fuse_fs *fs, const char *path, off_t off, int whence, + struct fuse_file_info *fi); +void fuse_fs_init(struct fuse_fs *fs, struct fuse_conn_info *conn, + struct fuse_config *cfg); +void fuse_fs_destroy(struct fuse_fs *fs); + +int fuse_notify_poll(struct fuse_pollhandle *ph); + +/** + * Create a new fuse filesystem object + * + * This is usually called from the factory of a fuse module to create + * a new instance of a filesystem. + * + * @param op the filesystem operations + * @param op_size the size of the fuse_operations structure + * @param private_data Initial value for the `private_data` + * field of `struct fuse_context`. May be overridden by the + * `struct fuse_operations.init` handler. + * @return a new filesystem object + */ +struct fuse_fs *fuse_fs_new(const struct fuse_operations *op, size_t op_size, + void *private_data); + +/** + * Factory for creating filesystem objects + * + * The function may use and remove options from 'args' that belong + * to this module. + * + * For now the 'fs' vector always contains exactly one filesystem. + * This is the filesystem which will be below the newly created + * filesystem in the stack. + * + * @param args the command line arguments + * @param fs NULL terminated filesystem object vector + * @return the new filesystem object + */ +typedef struct fuse_fs *(*fuse_module_factory_t)(struct fuse_args *args, + struct fuse_fs *fs[]); +/** + * Register filesystem module + * + * If the "-omodules=*name*_:..." option is present, filesystem + * objects are created and pushed onto the stack with the *factory_* + * function. + * + * @param name_ the name of this filesystem module + * @param factory_ the factory function for this filesystem module + */ +#define FUSE_REGISTER_MODULE(name_, factory_) \ + fuse_module_factory_t fuse_module_ ## name_ ## _factory = factory_ + +/** Get session from fuse object */ +struct fuse_session *fuse_get_session(struct fuse *f); + +/** + * Open a FUSE file descriptor and set up the mount for the given + * mountpoint and flags. + * + * @param mountpoint reference to the mount in the file system + * @param options mount options + * @return the FUSE file descriptor or -1 upon error + */ +int fuse_open_channel(const char *mountpoint, const char *options); + +#ifdef __cplusplus +} +#endif + +#endif /* FUSE_H_ */ diff --git a/tools/virtiofsd/fuse_common.h b/tools/virtiofsd/fuse_common.h new file mode 100644 index 0000000..2d686b2 --- /dev/null +++ b/tools/virtiofsd/fuse_common.h @@ -0,0 +1,823 @@ +/* FUSE: Filesystem in Userspace + Copyright (C) 2001-2007 Miklos Szeredi + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB. +*/ + +/** @file */ + +#if !defined(FUSE_H_) && !defined(FUSE_LOWLEVEL_H_) +#error "Never include directly; use or instead." +#endif + +#ifndef FUSE_COMMON_H_ +#define FUSE_COMMON_H_ + +#include "fuse_opt.h" +#include "fuse_log.h" +#include +#include + +/** Major version of FUSE library interface */ +#define FUSE_MAJOR_VERSION 3 + +/** Minor version of FUSE library interface */ +#define FUSE_MINOR_VERSION 2 + +#define FUSE_MAKE_VERSION(maj, min) ((maj) * 10 + (min)) +#define FUSE_VERSION FUSE_MAKE_VERSION(FUSE_MAJOR_VERSION, FUSE_MINOR_VERSION) + +#ifdef __cplusplus +extern "C" { +#endif + +/** + * Information about an open file. + * + * File Handles are created by the open, opendir, and create methods and closed + * by the release and releasedir methods. Multiple file handles may be + * concurrently open for the same file. Generally, a client will create one + * file handle per file descriptor, though in some cases multiple file + * descriptors can share a single file handle. + */ +struct fuse_file_info { + /** Open flags. Available in open() and release() */ + int flags; + + /** In case of a write operation indicates if this was caused + by a delayed write from the page cache. If so, then the + context's pid, uid, and gid fields will not be valid, and + the *fh* value may not match the *fh* value that would + have been sent with the corresponding individual write + requests if write caching had been disabled. */ + unsigned int writepage : 1; + + /** Can be filled in by open, to use direct I/O on this file. */ + unsigned int direct_io : 1; + + /** Can be filled in by open. It signals the kernel that any + currently cached file data (ie., data that the filesystem + provided the last time the file was open) need not be + invalidated. Has no effect when set in other contexts (in + particular it does nothing when set by opendir()). */ + unsigned int keep_cache : 1; + + /** Indicates a flush operation. Set in flush operation, also + maybe set in highlevel lock operation and lowlevel release + operation. */ + unsigned int flush : 1; + + /** Can be filled in by open, to indicate that the file is not + seekable. */ + unsigned int nonseekable : 1; + + /* Indicates that flock locks for this file should be + released. If set, lock_owner shall contain a valid value. + May only be set in ->release(). */ + unsigned int flock_release : 1; + + /** Can be filled in by opendir. It signals the kernel to + enable caching of entries returned by readdir(). Has no + effect when set in other contexts (in particular it does + nothing when set by open()). */ + unsigned int cache_readdir : 1; + + /** Padding. Reserved for future use*/ + unsigned int padding : 25; + unsigned int padding2 : 32; + + /** File handle id. May be filled in by filesystem in create, + * open, and opendir(). Available in most other file operations on the + * same file handle. */ + uint64_t fh; + + /** Lock owner id. Available in locking operations and flush */ + uint64_t lock_owner; + + /** Requested poll events. Available in ->poll. Only set on kernels + which support it. If unsupported, this field is set to zero. */ + uint32_t poll_events; +}; + +/** + * Configuration parameters passed to fuse_session_loop_mt() and + * fuse_loop_mt(). + */ +struct fuse_loop_config { + /** + * whether to use separate device fds for each thread + * (may increase performance) + */ + int clone_fd; + + /** + * The maximum number of available worker threads before they + * start to get deleted when they become idle. If not + * specified, the default is 10. + * + * Adjusting this has performance implications; a very small number + * of threads in the pool will cause a lot of thread creation and + * deletion overhead and performance may suffer. When set to 0, a new + * thread will be created to service every operation. + */ + unsigned int max_idle_threads; +}; + +/************************************************************************** + * Capability bits for 'fuse_conn_info.capable' and 'fuse_conn_info.want' * + **************************************************************************/ + +/** + * Indicates that the filesystem supports asynchronous read requests. + * + * If this capability is not requested/available, the kernel will + * ensure that there is at most one pending read request per + * file-handle at any time, and will attempt to order read requests by + * increasing offset. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_ASYNC_READ (1 << 0) + +/** + * Indicates that the filesystem supports "remote" locking. + * + * This feature is enabled by default when supported by the kernel, + * and if getlk() and setlk() handlers are implemented. + */ +#define FUSE_CAP_POSIX_LOCKS (1 << 1) + +/** + * Indicates that the filesystem supports the O_TRUNC open flag. If + * disabled, and an application specifies O_TRUNC, fuse first calls + * truncate() and then open() with O_TRUNC filtered out. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_ATOMIC_O_TRUNC (1 << 3) + +/** + * Indicates that the filesystem supports lookups of "." and "..". + * + * This feature is disabled by default. + */ +#define FUSE_CAP_EXPORT_SUPPORT (1 << 4) + +/** + * Indicates that the kernel should not apply the umask to the + * file mode on create operations. + * + * This feature is disabled by default. + */ +#define FUSE_CAP_DONT_MASK (1 << 6) + +/** + * Indicates that libfuse should try to use splice() when writing to + * the fuse device. This may improve performance. + * + * This feature is disabled by default. + */ +#define FUSE_CAP_SPLICE_WRITE (1 << 7) + +/** + * Indicates that libfuse should try to move pages instead of copying when + * writing to / reading from the fuse device. This may improve performance. + * + * This feature is disabled by default. + */ +#define FUSE_CAP_SPLICE_MOVE (1 << 8) + +/** + * Indicates that libfuse should try to use splice() when reading from + * the fuse device. This may improve performance. + * + * This feature is enabled by default when supported by the kernel and + * if the filesystem implements a write_buf() handler. + */ +#define FUSE_CAP_SPLICE_READ (1 << 9) + +/** + * If set, the calls to flock(2) will be emulated using POSIX locks and must + * then be handled by the filesystem's setlock() handler. + * + * If not set, flock(2) calls will be handled by the FUSE kernel module + * internally (so any access that does not go through the kernel cannot be taken + * into account). + * + * This feature is enabled by default when supported by the kernel and + * if the filesystem implements a flock() handler. + */ +#define FUSE_CAP_FLOCK_LOCKS (1 << 10) + +/** + * Indicates that the filesystem supports ioctl's on directories. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_IOCTL_DIR (1 << 11) + +/** + * Traditionally, while a file is open the FUSE kernel module only + * asks the filesystem for an update of the file's attributes when a + * client attempts to read beyond EOF. This is unsuitable for + * e.g. network filesystems, where the file contents may change + * without the kernel knowing about it. + * + * If this flag is set, FUSE will check the validity of the attributes + * on every read. If the attributes are no longer valid (i.e., if the + * *attr_timeout* passed to fuse_reply_attr() or set in `struct + * fuse_entry_param` has passed), it will first issue a `getattr` + * request. If the new mtime differs from the previous value, any + * cached file *contents* will be invalidated as well. + * + * This flag should always be set when available. If all file changes + * go through the kernel, *attr_timeout* should be set to a very large + * number to avoid unnecessary getattr() calls. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_AUTO_INVAL_DATA (1 << 12) + +/** + * Indicates that the filesystem supports readdirplus. + * + * This feature is enabled by default when supported by the kernel and if the + * filesystem implements a readdirplus() handler. + */ +#define FUSE_CAP_READDIRPLUS (1 << 13) + +/** + * Indicates that the filesystem supports adaptive readdirplus. + * + * If FUSE_CAP_READDIRPLUS is not set, this flag has no effect. + * + * If FUSE_CAP_READDIRPLUS is set and this flag is not set, the kernel + * will always issue readdirplus() requests to retrieve directory + * contents. + * + * If FUSE_CAP_READDIRPLUS is set and this flag is set, the kernel + * will issue both readdir() and readdirplus() requests, depending on + * how much information is expected to be required. + * + * As of Linux 4.20, the algorithm is as follows: when userspace + * starts to read directory entries, issue a READDIRPLUS request to + * the filesystem. If any entry attributes have been looked up by the + * time userspace requests the next batch of entries continue with + * READDIRPLUS, otherwise switch to plain READDIR. This will reasult + * in eg plain "ls" triggering READDIRPLUS first then READDIR after + * that because it doesn't do lookups. "ls -l" should result in all + * READDIRPLUS, except if dentries are already cached. + * + * This feature is enabled by default when supported by the kernel and + * if the filesystem implements both a readdirplus() and a readdir() + * handler. + */ +#define FUSE_CAP_READDIRPLUS_AUTO (1 << 14) + +/** + * Indicates that the filesystem supports asynchronous direct I/O submission. + * + * If this capability is not requested/available, the kernel will ensure that + * there is at most one pending read and one pending write request per direct + * I/O file-handle at any time. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_ASYNC_DIO (1 << 15) + +/** + * Indicates that writeback caching should be enabled. This means that + * individual write request may be buffered and merged in the kernel + * before they are send to the filesystem. + * + * This feature is disabled by default. + */ +#define FUSE_CAP_WRITEBACK_CACHE (1 << 16) + +/** + * Indicates support for zero-message opens. If this flag is set in + * the `capable` field of the `fuse_conn_info` structure, then the + * filesystem may return `ENOSYS` from the open() handler to indicate + * success. Further attempts to open files will be handled in the + * kernel. (If this flag is not set, returning ENOSYS will be treated + * as an error and signaled to the caller). + * + * Setting (or unsetting) this flag in the `want` field has *no + * effect*. + */ +#define FUSE_CAP_NO_OPEN_SUPPORT (1 << 17) + +/** + * Indicates support for parallel directory operations. If this flag + * is unset, the FUSE kernel module will ensure that lookup() and + * readdir() requests are never issued concurrently for the same + * directory. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_PARALLEL_DIROPS (1 << 18) + +/** + * Indicates support for POSIX ACLs. + * + * If this feature is enabled, the kernel will cache and have + * responsibility for enforcing ACLs. ACL will be stored as xattrs and + * passed to userspace, which is responsible for updating the ACLs in + * the filesystem, keeping the file mode in sync with the ACL, and + * ensuring inheritance of default ACLs when new filesystem nodes are + * created. Note that this requires that the file system is able to + * parse and interpret the xattr representation of ACLs. + * + * Enabling this feature implicitly turns on the + * ``default_permissions`` mount option (even if it was not passed to + * mount(2)). + * + * This feature is disabled by default. + */ +#define FUSE_CAP_POSIX_ACL (1 << 19) + +/** + * Indicates that the filesystem is responsible for unsetting + * setuid and setgid bits when a file is written, truncated, or + * its owner is changed. + * + * This feature is enabled by default when supported by the kernel. + */ +#define FUSE_CAP_HANDLE_KILLPRIV (1 << 20) + +/** + * Indicates support for zero-message opendirs. If this flag is set in + * the `capable` field of the `fuse_conn_info` structure, then the filesystem + * may return `ENOSYS` from the opendir() handler to indicate success. Further + * opendir and releasedir messages will be handled in the kernel. (If this + * flag is not set, returning ENOSYS will be treated as an error and signalled + * to the caller.) + * + * Setting (or unsetting) this flag in the `want` field has *no effect*. + */ +#define FUSE_CAP_NO_OPENDIR_SUPPORT (1 << 24) + +/** + * Ioctl flags + * + * FUSE_IOCTL_COMPAT: 32bit compat ioctl on 64bit machine + * FUSE_IOCTL_UNRESTRICTED: not restricted to well-formed ioctls, retry allowed + * FUSE_IOCTL_RETRY: retry with new iovecs + * FUSE_IOCTL_DIR: is a directory + * + * FUSE_IOCTL_MAX_IOV: maximum of in_iovecs + out_iovecs + */ +#define FUSE_IOCTL_COMPAT (1 << 0) +#define FUSE_IOCTL_UNRESTRICTED (1 << 1) +#define FUSE_IOCTL_RETRY (1 << 2) +#define FUSE_IOCTL_DIR (1 << 4) + +#define FUSE_IOCTL_MAX_IOV 256 + +/** + * Connection information, passed to the ->init() method + * + * Some of the elements are read-write, these can be changed to + * indicate the value requested by the filesystem. The requested + * value must usually be smaller than the indicated value. + */ +struct fuse_conn_info { + /** + * Major version of the protocol (read-only) + */ + unsigned proto_major; + + /** + * Minor version of the protocol (read-only) + */ + unsigned proto_minor; + + /** + * Maximum size of the write buffer + */ + unsigned max_write; + + /** + * Maximum size of read requests. A value of zero indicates no + * limit. However, even if the filesystem does not specify a + * limit, the maximum size of read requests will still be + * limited by the kernel. + * + * NOTE: For the time being, the maximum size of read requests + * must be set both here *and* passed to fuse_session_new() + * using the ``-o max_read=`` mount option. At some point + * in the future, specifying the mount option will no longer + * be necessary. + */ + unsigned max_read; + + /** + * Maximum readahead + */ + unsigned max_readahead; + + /** + * Capability flags that the kernel supports (read-only) + */ + unsigned capable; + + /** + * Capability flags that the filesystem wants to enable. + * + * libfuse attempts to initialize this field with + * reasonable default values before calling the init() handler. + */ + unsigned want; + + /** + * Maximum number of pending "background" requests. A + * background request is any type of request for which the + * total number is not limited by other means. As of kernel + * 4.8, only two types of requests fall into this category: + * + * 1. Read-ahead requests + * 2. Asynchronous direct I/O requests + * + * Read-ahead requests are generated (if max_readahead is + * non-zero) by the kernel to preemptively fill its caches + * when it anticipates that userspace will soon read more + * data. + * + * Asynchronous direct I/O requests are generated if + * FUSE_CAP_ASYNC_DIO is enabled and userspace submits a large + * direct I/O request. In this case the kernel will internally + * split it up into multiple smaller requests and submit them + * to the filesystem concurrently. + * + * Note that the following requests are *not* background + * requests: writeback requests (limited by the kernel's + * flusher algorithm), regular (i.e., synchronous and + * buffered) userspace read/write requests (limited to one per + * thread), asynchronous read requests (Linux's io_submit(2) + * call actually blocks, so these are also limited to one per + * thread). + */ + unsigned max_background; + + /** + * Kernel congestion threshold parameter. If the number of pending + * background requests exceeds this number, the FUSE kernel module will + * mark the filesystem as "congested". This instructs the kernel to + * expect that queued requests will take some time to complete, and to + * adjust its algorithms accordingly (e.g. by putting a waiting thread + * to sleep instead of using a busy-loop). + */ + unsigned congestion_threshold; + + /** + * When FUSE_CAP_WRITEBACK_CACHE is enabled, the kernel is responsible + * for updating mtime and ctime when write requests are received. The + * updated values are passed to the filesystem with setattr() requests. + * However, if the filesystem does not support the full resolution of + * the kernel timestamps (nanoseconds), the mtime and ctime values used + * by kernel and filesystem will differ (and result in an apparent + * change of times after a cache flush). + * + * To prevent this problem, this variable can be used to inform the + * kernel about the timestamp granularity supported by the file-system. + * The value should be power of 10. The default is 1, i.e. full + * nano-second resolution. Filesystems supporting only second resolution + * should set this to 1000000000. + */ + unsigned time_gran; + + /** + * For future use. + */ + unsigned reserved[22]; +}; + +struct fuse_session; +struct fuse_pollhandle; +struct fuse_conn_info_opts; + +/** + * This function parses several command-line options that can be used + * to override elements of struct fuse_conn_info. The pointer returned + * by this function should be passed to the + * fuse_apply_conn_info_opts() method by the file system's init() + * handler. + * + * Before using this function, think twice if you really want these + * parameters to be adjustable from the command line. In most cases, + * they should be determined by the file system internally. + * + * The following options are recognized: + * + * -o max_write=N sets conn->max_write + * -o max_readahead=N sets conn->max_readahead + * -o max_background=N sets conn->max_background + * -o congestion_threshold=N sets conn->congestion_threshold + * -o async_read sets FUSE_CAP_ASYNC_READ in conn->want + * -o sync_read unsets FUSE_CAP_ASYNC_READ in conn->want + * -o atomic_o_trunc sets FUSE_CAP_ATOMIC_O_TRUNC in conn->want + * -o no_remote_lock Equivalent to -o no_remote_flock,no_remote_posix_lock + * -o no_remote_flock Unsets FUSE_CAP_FLOCK_LOCKS in conn->want + * -o no_remote_posix_lock Unsets FUSE_CAP_POSIX_LOCKS in conn->want + * -o [no_]splice_write (un-)sets FUSE_CAP_SPLICE_WRITE in conn->want + * -o [no_]splice_move (un-)sets FUSE_CAP_SPLICE_MOVE in conn->want + * -o [no_]splice_read (un-)sets FUSE_CAP_SPLICE_READ in conn->want + * -o [no_]auto_inval_data (un-)sets FUSE_CAP_AUTO_INVAL_DATA in conn->want + * -o readdirplus=no unsets FUSE_CAP_READDIRPLUS in conn->want + * -o readdirplus=yes sets FUSE_CAP_READDIRPLUS and unsets + * FUSE_CAP_READDIRPLUS_AUTO in conn->want + * -o readdirplus=auto sets FUSE_CAP_READDIRPLUS and + * FUSE_CAP_READDIRPLUS_AUTO in conn->want + * -o [no_]async_dio (un-)sets FUSE_CAP_ASYNC_DIO in conn->want + * -o [no_]writeback_cache (un-)sets FUSE_CAP_WRITEBACK_CACHE in conn->want + * -o time_gran=N sets conn->time_gran + * + * Known options will be removed from *args*, unknown options will be + * passed through unchanged. + * + * @param args argument vector (input+output) + * @return parsed options + **/ +struct fuse_conn_info_opts* fuse_parse_conn_info_opts(struct fuse_args *args); + +/** + * This function applies the (parsed) parameters in *opts* to the + * *conn* pointer. It may modify the following fields: wants, + * max_write, max_readahead, congestion_threshold, max_background, + * time_gran. A field is only set (or unset) if the corresponding + * option has been explicitly set. + */ +void fuse_apply_conn_info_opts(struct fuse_conn_info_opts *opts, + struct fuse_conn_info *conn); + +/** + * Go into the background + * + * @param foreground if true, stay in the foreground + * @return 0 on success, -1 on failure + */ +int fuse_daemonize(int foreground); + +/** + * Get the version of the library + * + * @return the version + */ +int fuse_version(void); + +/** + * Get the full package version string of the library + * + * @return the package version + */ +const char *fuse_pkgversion(void); + +/** + * Destroy poll handle + * + * @param ph the poll handle + */ +void fuse_pollhandle_destroy(struct fuse_pollhandle *ph); + +/* ----------------------------------------------------------- * + * Data buffer * + * ----------------------------------------------------------- */ + +/** + * Buffer flags + */ +enum fuse_buf_flags { + /** + * Buffer contains a file descriptor + * + * If this flag is set, the .fd field is valid, otherwise the + * .mem fields is valid. + */ + FUSE_BUF_IS_FD = (1 << 1), + + /** + * Seek on the file descriptor + * + * If this flag is set then the .pos field is valid and is + * used to seek to the given offset before performing + * operation on file descriptor. + */ + FUSE_BUF_FD_SEEK = (1 << 2), + + /** + * Retry operation on file descriptor + * + * If this flag is set then retry operation on file descriptor + * until .size bytes have been copied or an error or EOF is + * detected. + */ + FUSE_BUF_FD_RETRY = (1 << 3), +}; + +/** + * Buffer copy flags + */ +enum fuse_buf_copy_flags { + /** + * Don't use splice(2) + * + * Always fall back to using read and write instead of + * splice(2) to copy data from one file descriptor to another. + * + * If this flag is not set, then only fall back if splice is + * unavailable. + */ + FUSE_BUF_NO_SPLICE = (1 << 1), + + /** + * Force splice + * + * Always use splice(2) to copy data from one file descriptor + * to another. If splice is not available, return -EINVAL. + */ + FUSE_BUF_FORCE_SPLICE = (1 << 2), + + /** + * Try to move data with splice. + * + * If splice is used, try to move pages from the source to the + * destination instead of copying. See documentation of + * SPLICE_F_MOVE in splice(2) man page. + */ + FUSE_BUF_SPLICE_MOVE = (1 << 3), + + /** + * Don't block on the pipe when copying data with splice + * + * Makes the operations on the pipe non-blocking (if the pipe + * is full or empty). See SPLICE_F_NONBLOCK in the splice(2) + * man page. + */ + FUSE_BUF_SPLICE_NONBLOCK= (1 << 4), +}; + +/** + * Single data buffer + * + * Generic data buffer for I/O, extended attributes, etc... Data may + * be supplied as a memory pointer or as a file descriptor + */ +struct fuse_buf { + /** + * Size of data in bytes + */ + size_t size; + + /** + * Buffer flags + */ + enum fuse_buf_flags flags; + + /** + * Memory pointer + * + * Used unless FUSE_BUF_IS_FD flag is set. + */ + void *mem; + + /** + * File descriptor + * + * Used if FUSE_BUF_IS_FD flag is set. + */ + int fd; + + /** + * File position + * + * Used if FUSE_BUF_FD_SEEK flag is set. + */ + off_t pos; +}; + +/** + * Data buffer vector + * + * An array of data buffers, each containing a memory pointer or a + * file descriptor. + * + * Allocate dynamically to add more than one buffer. + */ +struct fuse_bufvec { + /** + * Number of buffers in the array + */ + size_t count; + + /** + * Index of current buffer within the array + */ + size_t idx; + + /** + * Current offset within the current buffer + */ + size_t off; + + /** + * Array of buffers + */ + struct fuse_buf buf[1]; +}; + +/* Initialize bufvec with a single buffer of given size */ +#define FUSE_BUFVEC_INIT(size__) \ + ((struct fuse_bufvec) { \ + /* .count= */ 1, \ + /* .idx = */ 0, \ + /* .off = */ 0, \ + /* .buf = */ { /* [0] = */ { \ + /* .size = */ (size__), \ + /* .flags = */ (enum fuse_buf_flags) 0, \ + /* .mem = */ NULL, \ + /* .fd = */ -1, \ + /* .pos = */ 0, \ + } } \ + } ) + +/** + * Get total size of data in a fuse buffer vector + * + * @param bufv buffer vector + * @return size of data + */ +size_t fuse_buf_size(const struct fuse_bufvec *bufv); + +/** + * Copy data from one buffer vector to another + * + * @param dst destination buffer vector + * @param src source buffer vector + * @param flags flags controlling the copy + * @return actual number of bytes copied or -errno on error + */ +ssize_t fuse_buf_copy(struct fuse_bufvec *dst, struct fuse_bufvec *src, + enum fuse_buf_copy_flags flags); + +/* ----------------------------------------------------------- * + * Signal handling * + * ----------------------------------------------------------- */ + +/** + * Exit session on HUP, TERM and INT signals and ignore PIPE signal + * + * Stores session in a global variable. May only be called once per + * process until fuse_remove_signal_handlers() is called. + * + * Once either of the POSIX signals arrives, the signal handler calls + * fuse_session_exit(). + * + * @param se the session to exit + * @return 0 on success, -1 on failure + * + * See also: + * fuse_remove_signal_handlers() + */ +int fuse_set_signal_handlers(struct fuse_session *se); + +/** + * Restore default signal handlers + * + * Resets global session. After this fuse_set_signal_handlers() may + * be called again. + * + * @param se the same session as given in fuse_set_signal_handlers() + * + * See also: + * fuse_set_signal_handlers() + */ +void fuse_remove_signal_handlers(struct fuse_session *se); + +/* ----------------------------------------------------------- * + * Compatibility stuff * + * ----------------------------------------------------------- */ + +#if !defined(FUSE_USE_VERSION) || FUSE_USE_VERSION < 30 +# error only API version 30 or greater is supported +#endif + +#ifdef __cplusplus +} +#endif + + +/* + * This interface uses 64 bit off_t. + * + * On 32bit systems please add -D_FILE_OFFSET_BITS=64 to your compile flags! + */ + +#if defined(__GNUC__) && (__GNUC__ > 4 || __GNUC__ == 4 && __GNUC_MINOR__ >= 6) && !defined __cplusplus +_Static_assert(sizeof(off_t) == 8, "fuse: off_t must be 64bit"); +#else +struct _fuse_off_t_must_be_64bit_dummy_struct \ + { unsigned _fuse_off_t_must_be_64bit:((sizeof(off_t) == 8) ? 1 : -1); }; +#endif + +#endif /* FUSE_COMMON_H_ */ diff --git a/tools/virtiofsd/fuse_i.h b/tools/virtiofsd/fuse_i.h new file mode 100644 index 0000000..d38b630 --- /dev/null +++ b/tools/virtiofsd/fuse_i.h @@ -0,0 +1,139 @@ +/* + FUSE: Filesystem in Userspace + Copyright (C) 2001-2007 Miklos Szeredi + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB +*/ + +#include "fuse.h" +#include "fuse_lowlevel.h" + +struct mount_opts; + +struct fuse_req { + struct fuse_session *se; + uint64_t unique; + int ctr; + pthread_mutex_t lock; + struct fuse_ctx ctx; + struct fuse_chan *ch; + int interrupted; + unsigned int ioctl_64bit : 1; + union { + struct { + uint64_t unique; + } i; + struct { + fuse_interrupt_func_t func; + void *data; + } ni; + } u; + struct fuse_req *next; + struct fuse_req *prev; +}; + +struct fuse_notify_req { + uint64_t unique; + void (*reply)(struct fuse_notify_req *, fuse_req_t, fuse_ino_t, + const void *, const struct fuse_buf *); + struct fuse_notify_req *next; + struct fuse_notify_req *prev; +}; + +struct fuse_session { + char *mountpoint; + volatile int exited; + int fd; + struct mount_opts *mo; + int debug; + int deny_others; + struct fuse_lowlevel_ops op; + int got_init; + struct cuse_data *cuse_data; + void *userdata; + uid_t owner; + struct fuse_conn_info conn; + struct fuse_req list; + struct fuse_req interrupts; + pthread_mutex_t lock; + int got_destroy; + pthread_key_t pipe_key; + int broken_splice_nonblock; + uint64_t notify_ctr; + struct fuse_notify_req notify_list; + size_t bufsize; + int error; +}; + +struct fuse_chan { + pthread_mutex_t lock; + int ctr; + int fd; +}; + +/** + * Filesystem module + * + * Filesystem modules are registered with the FUSE_REGISTER_MODULE() + * macro. + * + */ +struct fuse_module { + char *name; + fuse_module_factory_t factory; + struct fuse_module *next; + struct fusemod_so *so; + int ctr; +}; + +/* ----------------------------------------------------------- * + * Channel interface (when using -o clone_fd) * + * ----------------------------------------------------------- */ + +/** + * Obtain counted reference to the channel + * + * @param ch the channel + * @return the channel + */ +struct fuse_chan *fuse_chan_get(struct fuse_chan *ch); + +/** + * Drop counted reference to a channel + * + * @param ch the channel + */ +void fuse_chan_put(struct fuse_chan *ch); + +struct mount_opts *parse_mount_opts(struct fuse_args *args); +void destroy_mount_opts(struct mount_opts *mo); +void fuse_mount_version(void); +unsigned get_max_read(struct mount_opts *o); +void fuse_kern_unmount(const char *mountpoint, int fd); +int fuse_kern_mount(const char *mountpoint, struct mount_opts *mo); + +int fuse_send_reply_iov_nofree(fuse_req_t req, int error, struct iovec *iov, + int count); +void fuse_free_req(fuse_req_t req); + +void cuse_lowlevel_init(fuse_req_t req, fuse_ino_t nodeide, const void *inarg); + +int fuse_start_thread(pthread_t *thread_id, void *(*func)(void *), void *arg); + +int fuse_session_receive_buf_int(struct fuse_session *se, struct fuse_buf *buf, + struct fuse_chan *ch); +void fuse_session_process_buf_int(struct fuse_session *se, + const struct fuse_buf *buf, struct fuse_chan *ch); + +struct fuse *fuse_new_31(struct fuse_args *args, const struct fuse_operations *op, + size_t op_size, void *private_data); +int fuse_loop_mt_32(struct fuse *f, struct fuse_loop_config *config); +int fuse_session_loop_mt_32(struct fuse_session *se, struct fuse_loop_config *config); + +#define FUSE_MAX_MAX_PAGES 256 +#define FUSE_DEFAULT_MAX_PAGES_PER_REQ 32 + +/* room needed in buffer to accommodate header */ +#define FUSE_BUFFER_HEADER_SIZE 0x1000 + diff --git a/tools/virtiofsd/fuse_log.h b/tools/virtiofsd/fuse_log.h new file mode 100644 index 0000000..5e112e0 --- /dev/null +++ b/tools/virtiofsd/fuse_log.h @@ -0,0 +1,82 @@ +/* + FUSE: Filesystem in Userspace + Copyright (C) 2019 Red Hat, Inc. + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB. +*/ + +#ifndef FUSE_LOG_H_ +#define FUSE_LOG_H_ + +/** @file + * + * This file defines the logging interface of FUSE + */ + +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/** + * Log severity level + * + * These levels correspond to syslog(2) log levels since they are widely used. + */ +enum fuse_log_level { + FUSE_LOG_EMERG, + FUSE_LOG_ALERT, + FUSE_LOG_CRIT, + FUSE_LOG_ERR, + FUSE_LOG_WARNING, + FUSE_LOG_NOTICE, + FUSE_LOG_INFO, + FUSE_LOG_DEBUG +}; + +/** + * Log message handler function. + * + * This function must be thread-safe. It may be called from any libfuse + * function, including fuse_parse_cmdline() and other functions invoked before + * a FUSE filesystem is created. + * + * Install a custom log message handler function using fuse_set_log_func(). + * + * @param level log severity level + * @param fmt sprintf-style format string including newline + * @param ap format string arguments + */ +typedef void (*fuse_log_func_t)(enum fuse_log_level level, + const char *fmt, va_list ap); + +/** + * Install a custom log handler function. + * + * Log messages are emitted by libfuse functions to report errors and debug + * information. Messages are printed to stderr by default but this can be + * overridden by installing a custom log message handler function. + * + * The log message handler function is global and affects all FUSE filesystems + * created within this process. + * + * @param func a custom log message handler function or NULL to revert to + * the default + */ +void fuse_set_log_func(fuse_log_func_t func); + +/** + * Emit a log message + * + * @param level severity level (FUSE_LOG_ERR, FUSE_LOG_DEBUG, etc) + * @param fmt sprintf-style format string including newline + */ +void fuse_log(enum fuse_log_level level, const char *fmt, ...); + +#ifdef __cplusplus +} +#endif + +#endif /* FUSE_LOG_H_ */ diff --git a/tools/virtiofsd/fuse_lowlevel.h b/tools/virtiofsd/fuse_lowlevel.h new file mode 100644 index 0000000..18c6363 --- /dev/null +++ b/tools/virtiofsd/fuse_lowlevel.h @@ -0,0 +1,2089 @@ +/* + FUSE: Filesystem in Userspace + Copyright (C) 2001-2007 Miklos Szeredi + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB. +*/ + +#ifndef FUSE_LOWLEVEL_H_ +#define FUSE_LOWLEVEL_H_ + +/** @file + * + * Low level API + * + * IMPORTANT: you should define FUSE_USE_VERSION before including this + * header. To use the newest API define it to 31 (recommended for any + * new application). + */ + +#ifndef FUSE_USE_VERSION +#error FUSE_USE_VERSION not defined +#endif + +#include "fuse_common.h" + +#include +#include +#include +#include +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +/* ----------------------------------------------------------- * + * Miscellaneous definitions * + * ----------------------------------------------------------- */ + +/** The node ID of the root inode */ +#define FUSE_ROOT_ID 1 + +/** Inode number type */ +typedef uint64_t fuse_ino_t; + +/** Request pointer type */ +typedef struct fuse_req *fuse_req_t; + +/** + * Session + * + * This provides hooks for processing requests, and exiting + */ +struct fuse_session; + +/** Directory entry parameters supplied to fuse_reply_entry() */ +struct fuse_entry_param { + /** Unique inode number + * + * In lookup, zero means negative entry (from version 2.5) + * Returning ENOENT also means negative entry, but by setting zero + * ino the kernel may cache negative entries for entry_timeout + * seconds. + */ + fuse_ino_t ino; + + /** Generation number for this entry. + * + * If the file system will be exported over NFS, the + * ino/generation pairs need to be unique over the file + * system's lifetime (rather than just the mount time). So if + * the file system reuses an inode after it has been deleted, + * it must assign a new, previously unused generation number + * to the inode at the same time. + * + */ + uint64_t generation; + + /** Inode attributes. + * + * Even if attr_timeout == 0, attr must be correct. For example, + * for open(), FUSE uses attr.st_size from lookup() to determine + * how many bytes to request. If this value is not correct, + * incorrect data will be returned. + */ + struct stat attr; + + /** Validity timeout (in seconds) for inode attributes. If + attributes only change as a result of requests that come + through the kernel, this should be set to a very large + value. */ + double attr_timeout; + + /** Validity timeout (in seconds) for the name. If directory + entries are changed/deleted only as a result of requests + that come through the kernel, this should be set to a very + large value. */ + double entry_timeout; +}; + +/** + * Additional context associated with requests. + * + * Note that the reported client uid, gid and pid may be zero in some + * situations. For example, if the FUSE file system is running in a + * PID or user namespace but then accessed from outside the namespace, + * there is no valid uid/pid/gid that could be reported. + */ +struct fuse_ctx { + /** User ID of the calling process */ + uid_t uid; + + /** Group ID of the calling process */ + gid_t gid; + + /** Thread ID of the calling process */ + pid_t pid; + + /** Umask of the calling process */ + mode_t umask; +}; + +struct fuse_forget_data { + fuse_ino_t ino; + uint64_t nlookup; +}; + +/* 'to_set' flags in setattr */ +#define FUSE_SET_ATTR_MODE (1 << 0) +#define FUSE_SET_ATTR_UID (1 << 1) +#define FUSE_SET_ATTR_GID (1 << 2) +#define FUSE_SET_ATTR_SIZE (1 << 3) +#define FUSE_SET_ATTR_ATIME (1 << 4) +#define FUSE_SET_ATTR_MTIME (1 << 5) +#define FUSE_SET_ATTR_ATIME_NOW (1 << 7) +#define FUSE_SET_ATTR_MTIME_NOW (1 << 8) +#define FUSE_SET_ATTR_CTIME (1 << 10) + +/* ----------------------------------------------------------- * + * Request methods and replies * + * ----------------------------------------------------------- */ + +/** + * Low level filesystem operations + * + * Most of the methods (with the exception of init and destroy) + * receive a request handle (fuse_req_t) as their first argument. + * This handle must be passed to one of the specified reply functions. + * + * This may be done inside the method invocation, or after the call + * has returned. The request handle is valid until one of the reply + * functions is called. + * + * Other pointer arguments (name, fuse_file_info, etc) are not valid + * after the call has returned, so if they are needed later, their + * contents have to be copied. + * + * In general, all methods are expected to perform any necessary + * permission checking. However, a filesystem may delegate this task + * to the kernel by passing the `default_permissions` mount option to + * `fuse_session_new()`. In this case, methods will only be called if + * the kernel's permission check has succeeded. + * + * The filesystem sometimes needs to handle a return value of -ENOENT + * from the reply function, which means, that the request was + * interrupted, and the reply discarded. For example if + * fuse_reply_open() return -ENOENT means, that the release method for + * this file will not be called. + */ +struct fuse_lowlevel_ops { + /** + * Initialize filesystem + * + * This function is called when libfuse establishes + * communication with the FUSE kernel module. The file system + * should use this module to inspect and/or modify the + * connection parameters provided in the `conn` structure. + * + * Note that some parameters may be overwritten by options + * passed to fuse_session_new() which take precedence over the + * values set in this handler. + * + * There's no reply to this function + * + * @param userdata the user data passed to fuse_session_new() + */ + void (*init) (void *userdata, struct fuse_conn_info *conn); + + /** + * Clean up filesystem. + * + * Called on filesystem exit. When this method is called, the + * connection to the kernel may be gone already, so that eg. calls + * to fuse_lowlevel_notify_* will fail. + * + * There's no reply to this function + * + * @param userdata the user data passed to fuse_session_new() + */ + void (*destroy) (void *userdata); + + /** + * Look up a directory entry by name and get its attributes. + * + * Valid replies: + * fuse_reply_entry + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the parent directory + * @param name the name to look up + */ + void (*lookup) (fuse_req_t req, fuse_ino_t parent, const char *name); + + /** + * Forget about an inode + * + * This function is called when the kernel removes an inode + * from its internal caches. + * + * The inode's lookup count increases by one for every call to + * fuse_reply_entry and fuse_reply_create. The nlookup parameter + * indicates by how much the lookup count should be decreased. + * + * Inodes with a non-zero lookup count may receive request from + * the kernel even after calls to unlink, rmdir or (when + * overwriting an existing file) rename. Filesystems must handle + * such requests properly and it is recommended to defer removal + * of the inode until the lookup count reaches zero. Calls to + * unlink, rmdir or rename will be followed closely by forget + * unless the file or directory is open, in which case the + * kernel issues forget only after the release or releasedir + * calls. + * + * Note that if a file system will be exported over NFS the + * inodes lifetime must extend even beyond forget. See the + * generation field in struct fuse_entry_param above. + * + * On unmount the lookup count for all inodes implicitly drops + * to zero. It is not guaranteed that the file system will + * receive corresponding forget messages for the affected + * inodes. + * + * Valid replies: + * fuse_reply_none + * + * @param req request handle + * @param ino the inode number + * @param nlookup the number of lookups to forget + */ + void (*forget) (fuse_req_t req, fuse_ino_t ino, uint64_t nlookup); + + /** + * Get file attributes. + * + * If writeback caching is enabled, the kernel may have a + * better idea of a file's length than the FUSE file system + * (eg if there has been a write that extended the file size, + * but that has not yet been passed to the filesystem.n + * + * In this case, the st_size value provided by the file system + * will be ignored. + * + * Valid replies: + * fuse_reply_attr + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi for future use, currently always NULL + */ + void (*getattr) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi); + + /** + * Set file attributes + * + * In the 'attr' argument only members indicated by the 'to_set' + * bitmask contain valid values. Other members contain undefined + * values. + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits if the file + * size or owner is being changed. + * + * If the setattr was invoked from the ftruncate() system call + * under Linux kernel versions 2.6.15 or later, the fi->fh will + * contain the value set by the open method or will be undefined + * if the open method didn't set any value. Otherwise (not + * ftruncate call, or kernel version earlier than 2.6.15) the fi + * parameter will be NULL. + * + * Valid replies: + * fuse_reply_attr + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param attr the attributes + * @param to_set bit mask of attributes which should be set + * @param fi file information, or NULL + */ + void (*setattr) (fuse_req_t req, fuse_ino_t ino, struct stat *attr, + int to_set, struct fuse_file_info *fi); + + /** + * Read symbolic link + * + * Valid replies: + * fuse_reply_readlink + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + */ + void (*readlink) (fuse_req_t req, fuse_ino_t ino); + + /** + * Create file node + * + * Create a regular file, character device, block device, fifo or + * socket node. + * + * Valid replies: + * fuse_reply_entry + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the parent directory + * @param name to create + * @param mode file type and mode with which to create the new file + * @param rdev the device number (only valid if created file is a device) + */ + void (*mknod) (fuse_req_t req, fuse_ino_t parent, const char *name, + mode_t mode, dev_t rdev); + + /** + * Create a directory + * + * Valid replies: + * fuse_reply_entry + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the parent directory + * @param name to create + * @param mode with which to create the new file + */ + void (*mkdir) (fuse_req_t req, fuse_ino_t parent, const char *name, + mode_t mode); + + /** + * Remove a file + * + * If the file's inode's lookup count is non-zero, the file + * system is expected to postpone any removal of the inode + * until the lookup count reaches zero (see description of the + * forget function). + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the parent directory + * @param name to remove + */ + void (*unlink) (fuse_req_t req, fuse_ino_t parent, const char *name); + + /** + * Remove a directory + * + * If the directory's inode's lookup count is non-zero, the + * file system is expected to postpone any removal of the + * inode until the lookup count reaches zero (see description + * of the forget function). + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the parent directory + * @param name to remove + */ + void (*rmdir) (fuse_req_t req, fuse_ino_t parent, const char *name); + + /** + * Create a symbolic link + * + * Valid replies: + * fuse_reply_entry + * fuse_reply_err + * + * @param req request handle + * @param link the contents of the symbolic link + * @param parent inode number of the parent directory + * @param name to create + */ + void (*symlink) (fuse_req_t req, const char *link, fuse_ino_t parent, + const char *name); + + /** Rename a file + * + * If the target exists it should be atomically replaced. If + * the target's inode's lookup count is non-zero, the file + * system is expected to postpone any removal of the inode + * until the lookup count reaches zero (see description of the + * forget function). + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EINVAL, i.e. all + * future bmap requests will fail with EINVAL without being + * send to the filesystem process. + * + * *flags* may be `RENAME_EXCHANGE` or `RENAME_NOREPLACE`. If + * RENAME_NOREPLACE is specified, the filesystem must not + * overwrite *newname* if it exists and return an error + * instead. If `RENAME_EXCHANGE` is specified, the filesystem + * must atomically exchange the two files, i.e. both must + * exist and neither may be deleted. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the old parent directory + * @param name old name + * @param newparent inode number of the new parent directory + * @param newname new name + */ + void (*rename) (fuse_req_t req, fuse_ino_t parent, const char *name, + fuse_ino_t newparent, const char *newname, + unsigned int flags); + + /** + * Create a hard link + * + * Valid replies: + * fuse_reply_entry + * fuse_reply_err + * + * @param req request handle + * @param ino the old inode number + * @param newparent inode number of the new parent directory + * @param newname new name to create + */ + void (*link) (fuse_req_t req, fuse_ino_t ino, fuse_ino_t newparent, + const char *newname); + + /** + * Open a file + * + * Open flags are available in fi->flags. The following rules + * apply. + * + * - Creation (O_CREAT, O_EXCL, O_NOCTTY) flags will be + * filtered out / handled by the kernel. + * + * - Access modes (O_RDONLY, O_WRONLY, O_RDWR) should be used + * by the filesystem to check if the operation is + * permitted. If the ``-o default_permissions`` mount + * option is given, this check is already done by the + * kernel before calling open() and may thus be omitted by + * the filesystem. + * + * - When writeback caching is enabled, the kernel may send + * read requests even for files opened with O_WRONLY. The + * filesystem should be prepared to handle this. + * + * - When writeback caching is disabled, the filesystem is + * expected to properly handle the O_APPEND flag and ensure + * that each write is appending to the end of the file. + * + * - When writeback caching is enabled, the kernel will + * handle O_APPEND. However, unless all changes to the file + * come through the kernel this will not work reliably. The + * filesystem should thus either ignore the O_APPEND flag + * (and let the kernel handle it), or return an error + * (indicating that reliably O_APPEND is not available). + * + * Filesystem may store an arbitrary file handle (pointer, + * index, etc) in fi->fh, and use this in other all other file + * operations (read, write, flush, release, fsync). + * + * Filesystem may also implement stateless file I/O and not store + * anything in fi->fh. + * + * There are also some flags (direct_io, keep_cache) which the + * filesystem may set in fi, to change the way the file is opened. + * See fuse_file_info structure in for more details. + * + * If this request is answered with an error code of ENOSYS + * and FUSE_CAP_NO_OPEN_SUPPORT is set in + * `fuse_conn_info.capable`, this is treated as success and + * future calls to open and release will also succeed without being + * sent to the filesystem process. + * + * Valid replies: + * fuse_reply_open + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + */ + void (*open) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi); + + /** + * Read data + * + * Read should send exactly the number of bytes requested except + * on EOF or error, otherwise the rest of the data will be + * substituted with zeroes. An exception to this is when the file + * has been opened in 'direct_io' mode, in which case the return + * value of the read system call will reflect the return value of + * this operation. + * + * fi->fh will contain the value set by the open method, or will + * be undefined if the open method didn't set any value. + * + * Valid replies: + * fuse_reply_buf + * fuse_reply_iov + * fuse_reply_data + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param size number of bytes to read + * @param off offset to read from + * @param fi file information + */ + void (*read) (fuse_req_t req, fuse_ino_t ino, size_t size, off_t off, + struct fuse_file_info *fi); + + /** + * Write data + * + * Write should return exactly the number of bytes requested + * except on error. An exception to this is when the file has + * been opened in 'direct_io' mode, in which case the return value + * of the write system call will reflect the return value of this + * operation. + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits. + * + * fi->fh will contain the value set by the open method, or will + * be undefined if the open method didn't set any value. + * + * Valid replies: + * fuse_reply_write + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param buf data to write + * @param size number of bytes to write + * @param off offset to write to + * @param fi file information + */ + void (*write) (fuse_req_t req, fuse_ino_t ino, const char *buf, + size_t size, off_t off, struct fuse_file_info *fi); + + /** + * Flush method + * + * This is called on each close() of the opened file. + * + * Since file descriptors can be duplicated (dup, dup2, fork), for + * one open call there may be many flush calls. + * + * Filesystems shouldn't assume that flush will always be called + * after some writes, or that if will be called at all. + * + * fi->fh will contain the value set by the open method, or will + * be undefined if the open method didn't set any value. + * + * NOTE: the name of the method is misleading, since (unlike + * fsync) the filesystem is not forced to flush pending writes. + * One reason to flush data is if the filesystem wants to return + * write errors during close. However, such use is non-portable + * because POSIX does not require [close] to wait for delayed I/O to + * complete. + * + * If the filesystem supports file locking operations (setlk, + * getlk) it should remove all locks belonging to 'fi->owner'. + * + * If this request is answered with an error code of ENOSYS, + * this is treated as success and future calls to flush() will + * succeed automatically without being send to the filesystem + * process. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + * + * [close]: http://pubs.opengroup.org/onlinepubs/9699919799/functions/close.html + */ + void (*flush) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi); + + /** + * Release an open file + * + * Release is called when there are no more references to an open + * file: all file descriptors are closed and all memory mappings + * are unmapped. + * + * For every open call there will be exactly one release call (unless + * the filesystem is force-unmounted). + * + * The filesystem may reply with an error, but error values are + * not returned to close() or munmap() which triggered the + * release. + * + * fi->fh will contain the value set by the open method, or will + * be undefined if the open method didn't set any value. + * fi->flags will contain the same flags as for open. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + */ + void (*release) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi); + + /** + * Synchronize file contents + * + * If the datasync parameter is non-zero, then only the user data + * should be flushed, not the meta data. + * + * If this request is answered with an error code of ENOSYS, + * this is treated as success and future calls to fsync() will + * succeed automatically without being send to the filesystem + * process. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param datasync flag indicating if only data should be flushed + * @param fi file information + */ + void (*fsync) (fuse_req_t req, fuse_ino_t ino, int datasync, + struct fuse_file_info *fi); + + /** + * Open a directory + * + * Filesystem may store an arbitrary file handle (pointer, index, + * etc) in fi->fh, and use this in other all other directory + * stream operations (readdir, releasedir, fsyncdir). + * + * If this request is answered with an error code of ENOSYS and + * FUSE_CAP_NO_OPENDIR_SUPPORT is set in `fuse_conn_info.capable`, + * this is treated as success and future calls to opendir and + * releasedir will also succeed without being sent to the filesystem + * process. In addition, the kernel will cache readdir results + * as if opendir returned FOPEN_KEEP_CACHE | FOPEN_CACHE_DIR. + * + * Valid replies: + * fuse_reply_open + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + */ + void (*opendir) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi); + + /** + * Read directory + * + * Send a buffer filled using fuse_add_direntry(), with size not + * exceeding the requested size. Send an empty buffer on end of + * stream. + * + * fi->fh will contain the value set by the opendir method, or + * will be undefined if the opendir method didn't set any value. + * + * Returning a directory entry from readdir() does not affect + * its lookup count. + * + * If off_t is non-zero, then it will correspond to one of the off_t + * values that was previously returned by readdir() for the same + * directory handle. In this case, readdir() should skip over entries + * coming before the position defined by the off_t value. If entries + * are added or removed while the directory handle is open, they filesystem + * may still include the entries that have been removed, and may not + * report the entries that have been created. However, addition or + * removal of entries must never cause readdir() to skip over unrelated + * entries or to report them more than once. This means + * that off_t can not be a simple index that enumerates the entries + * that have been returned but must contain sufficient information to + * uniquely determine the next directory entry to return even when the + * set of entries is changing. + * + * The function does not have to report the '.' and '..' + * entries, but is allowed to do so. Note that, if readdir does + * not return '.' or '..', they will not be implicitly returned, + * and this behavior is observable by the caller. + * + * Valid replies: + * fuse_reply_buf + * fuse_reply_data + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param size maximum number of bytes to send + * @param off offset to continue reading the directory stream + * @param fi file information + */ + void (*readdir) (fuse_req_t req, fuse_ino_t ino, size_t size, off_t off, + struct fuse_file_info *fi); + + /** + * Release an open directory + * + * For every opendir call there will be exactly one releasedir + * call (unless the filesystem is force-unmounted). + * + * fi->fh will contain the value set by the opendir method, or + * will be undefined if the opendir method didn't set any value. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + */ + void (*releasedir) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi); + + /** + * Synchronize directory contents + * + * If the datasync parameter is non-zero, then only the directory + * contents should be flushed, not the meta data. + * + * fi->fh will contain the value set by the opendir method, or + * will be undefined if the opendir method didn't set any value. + * + * If this request is answered with an error code of ENOSYS, + * this is treated as success and future calls to fsyncdir() will + * succeed automatically without being send to the filesystem + * process. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param datasync flag indicating if only data should be flushed + * @param fi file information + */ + void (*fsyncdir) (fuse_req_t req, fuse_ino_t ino, int datasync, + struct fuse_file_info *fi); + + /** + * Get file system statistics + * + * Valid replies: + * fuse_reply_statfs + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number, zero means "undefined" + */ + void (*statfs) (fuse_req_t req, fuse_ino_t ino); + + /** + * Set an extended attribute + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EOPNOTSUPP, i.e. all + * future setxattr() requests will fail with EOPNOTSUPP without being + * send to the filesystem process. + * + * Valid replies: + * fuse_reply_err + */ + void (*setxattr) (fuse_req_t req, fuse_ino_t ino, const char *name, + const char *value, size_t size, int flags); + + /** + * Get an extended attribute + * + * If size is zero, the size of the value should be sent with + * fuse_reply_xattr. + * + * If the size is non-zero, and the value fits in the buffer, the + * value should be sent with fuse_reply_buf. + * + * If the size is too small for the value, the ERANGE error should + * be sent. + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EOPNOTSUPP, i.e. all + * future getxattr() requests will fail with EOPNOTSUPP without being + * send to the filesystem process. + * + * Valid replies: + * fuse_reply_buf + * fuse_reply_data + * fuse_reply_xattr + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param name of the extended attribute + * @param size maximum size of the value to send + */ + void (*getxattr) (fuse_req_t req, fuse_ino_t ino, const char *name, + size_t size); + + /** + * List extended attribute names + * + * If size is zero, the total size of the attribute list should be + * sent with fuse_reply_xattr. + * + * If the size is non-zero, and the null character separated + * attribute list fits in the buffer, the list should be sent with + * fuse_reply_buf. + * + * If the size is too small for the list, the ERANGE error should + * be sent. + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EOPNOTSUPP, i.e. all + * future listxattr() requests will fail with EOPNOTSUPP without being + * send to the filesystem process. + * + * Valid replies: + * fuse_reply_buf + * fuse_reply_data + * fuse_reply_xattr + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param size maximum size of the list to send + */ + void (*listxattr) (fuse_req_t req, fuse_ino_t ino, size_t size); + + /** + * Remove an extended attribute + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EOPNOTSUPP, i.e. all + * future removexattr() requests will fail with EOPNOTSUPP without being + * send to the filesystem process. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param name of the extended attribute + */ + void (*removexattr) (fuse_req_t req, fuse_ino_t ino, const char *name); + + /** + * Check file access permissions + * + * This will be called for the access() and chdir() system + * calls. If the 'default_permissions' mount option is given, + * this method is not called. + * + * This method is not called under Linux kernel versions 2.4.x + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent success, i.e. this and all future access() + * requests will succeed without being send to the filesystem process. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param mask requested access mode + */ + void (*access) (fuse_req_t req, fuse_ino_t ino, int mask); + + /** + * Create and open a file + * + * If the file does not exist, first create it with the specified + * mode, and then open it. + * + * See the description of the open handler for more + * information. + * + * If this method is not implemented or under Linux kernel + * versions earlier than 2.6.15, the mknod() and open() methods + * will be called instead. + * + * If this request is answered with an error code of ENOSYS, the handler + * is treated as not implemented (i.e., for this and future requests the + * mknod() and open() handlers will be called instead). + * + * Valid replies: + * fuse_reply_create + * fuse_reply_err + * + * @param req request handle + * @param parent inode number of the parent directory + * @param name to create + * @param mode file type and mode with which to create the new file + * @param fi file information + */ + void (*create) (fuse_req_t req, fuse_ino_t parent, const char *name, + mode_t mode, struct fuse_file_info *fi); + + /** + * Test for a POSIX file lock + * + * Valid replies: + * fuse_reply_lock + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + * @param lock the region/type to test + */ + void (*getlk) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi, struct flock *lock); + + /** + * Acquire, modify or release a POSIX file lock + * + * For POSIX threads (NPTL) there's a 1-1 relation between pid and + * owner, but otherwise this is not always the case. For checking + * lock ownership, 'fi->owner' must be used. The l_pid field in + * 'struct flock' should only be used to fill in this field in + * getlk(). + * + * Note: if the locking methods are not implemented, the kernel + * will still allow file locking to work locally. Hence these are + * only interesting for network filesystems and similar. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + * @param lock the region/type to set + * @param sleep locking operation may sleep + */ + void (*setlk) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi, + struct flock *lock, int sleep); + + /** + * Map block index within file to block index within device + * + * Note: This makes sense only for block device backed filesystems + * mounted with the 'blkdev' option + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure, i.e. all future bmap() requests will + * fail with the same error code without being send to the filesystem + * process. + * + * Valid replies: + * fuse_reply_bmap + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param blocksize unit of block index + * @param idx block index within file + */ + void (*bmap) (fuse_req_t req, fuse_ino_t ino, size_t blocksize, + uint64_t idx); + + /** + * Ioctl + * + * Note: For unrestricted ioctls (not allowed for FUSE + * servers), data in and out areas can be discovered by giving + * iovs and setting FUSE_IOCTL_RETRY in *flags*. For + * restricted ioctls, kernel prepares in/out data area + * according to the information encoded in cmd. + * + * Valid replies: + * fuse_reply_ioctl_retry + * fuse_reply_ioctl + * fuse_reply_ioctl_iov + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param cmd ioctl command + * @param arg ioctl argument + * @param fi file information + * @param flags for FUSE_IOCTL_* flags + * @param in_buf data fetched from the caller + * @param in_bufsz number of fetched bytes + * @param out_bufsz maximum size of output data + * + * Note : the unsigned long request submitted by the application + * is truncated to 32 bits. + */ + void (*ioctl) (fuse_req_t req, fuse_ino_t ino, unsigned int cmd, + void *arg, struct fuse_file_info *fi, unsigned flags, + const void *in_buf, size_t in_bufsz, size_t out_bufsz); + + /** + * Poll for IO readiness + * + * Note: If ph is non-NULL, the client should notify + * when IO readiness events occur by calling + * fuse_lowlevel_notify_poll() with the specified ph. + * + * Regardless of the number of times poll with a non-NULL ph + * is received, single notification is enough to clear all. + * Notifying more times incurs overhead but doesn't harm + * correctness. + * + * The callee is responsible for destroying ph with + * fuse_pollhandle_destroy() when no longer in use. + * + * If this request is answered with an error code of ENOSYS, this is + * treated as success (with a kernel-defined default poll-mask) and + * future calls to pull() will succeed the same way without being send + * to the filesystem process. + * + * Valid replies: + * fuse_reply_poll + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + * @param ph poll handle to be used for notification + */ + void (*poll) (fuse_req_t req, fuse_ino_t ino, struct fuse_file_info *fi, + struct fuse_pollhandle *ph); + + /** + * Write data made available in a buffer + * + * This is a more generic version of the ->write() method. If + * FUSE_CAP_SPLICE_READ is set in fuse_conn_info.want and the + * kernel supports splicing from the fuse device, then the + * data will be made available in pipe for supporting zero + * copy data transfer. + * + * buf->count is guaranteed to be one (and thus buf->idx is + * always zero). The write_buf handler must ensure that + * bufv->off is correctly updated (reflecting the number of + * bytes read from bufv->buf[0]). + * + * Unless FUSE_CAP_HANDLE_KILLPRIV is disabled, this method is + * expected to reset the setuid and setgid bits. + * + * Valid replies: + * fuse_reply_write + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param bufv buffer containing the data + * @param off offset to write to + * @param fi file information + */ + void (*write_buf) (fuse_req_t req, fuse_ino_t ino, + struct fuse_bufvec *bufv, off_t off, + struct fuse_file_info *fi); + + /** + * Callback function for the retrieve request + * + * Valid replies: + * fuse_reply_none + * + * @param req request handle + * @param cookie user data supplied to fuse_lowlevel_notify_retrieve() + * @param ino the inode number supplied to fuse_lowlevel_notify_retrieve() + * @param offset the offset supplied to fuse_lowlevel_notify_retrieve() + * @param bufv the buffer containing the returned data + */ + void (*retrieve_reply) (fuse_req_t req, void *cookie, fuse_ino_t ino, + off_t offset, struct fuse_bufvec *bufv); + + /** + * Forget about multiple inodes + * + * See description of the forget function for more + * information. + * + * Valid replies: + * fuse_reply_none + * + * @param req request handle + */ + void (*forget_multi) (fuse_req_t req, size_t count, + struct fuse_forget_data *forgets); + + /** + * Acquire, modify or release a BSD file lock + * + * Note: if the locking methods are not implemented, the kernel + * will still allow file locking to work locally. Hence these are + * only interesting for network filesystems and similar. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param fi file information + * @param op the locking operation, see flock(2) + */ + void (*flock) (fuse_req_t req, fuse_ino_t ino, + struct fuse_file_info *fi, int op); + + /** + * Allocate requested space. If this function returns success then + * subsequent writes to the specified range shall not fail due to the lack + * of free space on the file system storage media. + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EOPNOTSUPP, i.e. all + * future fallocate() requests will fail with EOPNOTSUPP without being + * send to the filesystem process. + * + * Valid replies: + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param offset starting point for allocated region + * @param length size of allocated region + * @param mode determines the operation to be performed on the given range, + * see fallocate(2) + */ + void (*fallocate) (fuse_req_t req, fuse_ino_t ino, int mode, + off_t offset, off_t length, struct fuse_file_info *fi); + + /** + * Read directory with attributes + * + * Send a buffer filled using fuse_add_direntry_plus(), with size not + * exceeding the requested size. Send an empty buffer on end of + * stream. + * + * fi->fh will contain the value set by the opendir method, or + * will be undefined if the opendir method didn't set any value. + * + * In contrast to readdir() (which does not affect the lookup counts), + * the lookup count of every entry returned by readdirplus(), except "." + * and "..", is incremented by one. + * + * Valid replies: + * fuse_reply_buf + * fuse_reply_data + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param size maximum number of bytes to send + * @param off offset to continue reading the directory stream + * @param fi file information + */ + void (*readdirplus) (fuse_req_t req, fuse_ino_t ino, size_t size, off_t off, + struct fuse_file_info *fi); + + /** + * Copy a range of data from one file to another + * + * Performs an optimized copy between two file descriptors without the + * additional cost of transferring data through the FUSE kernel module + * to user space (glibc) and then back into the FUSE filesystem again. + * + * In case this method is not implemented, glibc falls back to reading + * data from the source and writing to the destination. Effectively + * doing an inefficient copy of the data. + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure with error code EOPNOTSUPP, i.e. all + * future copy_file_range() requests will fail with EOPNOTSUPP without + * being send to the filesystem process. + * + * Valid replies: + * fuse_reply_write + * fuse_reply_err + * + * @param req request handle + * @param ino_in the inode number or the source file + * @param off_in starting point from were the data should be read + * @param fi_in file information of the source file + * @param ino_out the inode number or the destination file + * @param off_out starting point where the data should be written + * @param fi_out file information of the destination file + * @param len maximum size of the data to copy + * @param flags passed along with the copy_file_range() syscall + */ + void (*copy_file_range) (fuse_req_t req, fuse_ino_t ino_in, + off_t off_in, struct fuse_file_info *fi_in, + fuse_ino_t ino_out, off_t off_out, + struct fuse_file_info *fi_out, size_t len, + int flags); + + /** + * Find next data or hole after the specified offset + * + * If this request is answered with an error code of ENOSYS, this is + * treated as a permanent failure, i.e. all future lseek() requests will + * fail with the same error code without being send to the filesystem + * process. + * + * Valid replies: + * fuse_reply_lseek + * fuse_reply_err + * + * @param req request handle + * @param ino the inode number + * @param off offset to start search from + * @param whence either SEEK_DATA or SEEK_HOLE + * @param fi file information + */ + void (*lseek) (fuse_req_t req, fuse_ino_t ino, off_t off, int whence, + struct fuse_file_info *fi); +}; + +/** + * Reply with an error code or success. + * + * Possible requests: + * all except forget + * + * Whereever possible, error codes should be chosen from the list of + * documented error conditions in the corresponding system calls + * manpage. + * + * An error code of ENOSYS is sometimes treated specially. This is + * indicated in the documentation of the affected handler functions. + * + * The following requests may be answered with a zero error code: + * unlink, rmdir, rename, flush, release, fsync, fsyncdir, setxattr, + * removexattr, setlk. + * + * @param req request handle + * @param err the positive error value, or zero for success + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_err(fuse_req_t req, int err); + +/** + * Don't send reply + * + * Possible requests: + * forget + * forget_multi + * retrieve_reply + * + * @param req request handle + */ +void fuse_reply_none(fuse_req_t req); + +/** + * Reply with a directory entry + * + * Possible requests: + * lookup, mknod, mkdir, symlink, link + * + * Side effects: + * increments the lookup count on success + * + * @param req request handle + * @param e the entry parameters + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_entry(fuse_req_t req, const struct fuse_entry_param *e); + +/** + * Reply with a directory entry and open parameters + * + * currently the following members of 'fi' are used: + * fh, direct_io, keep_cache + * + * Possible requests: + * create + * + * Side effects: + * increments the lookup count on success + * + * @param req request handle + * @param e the entry parameters + * @param fi file information + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_create(fuse_req_t req, const struct fuse_entry_param *e, + const struct fuse_file_info *fi); + +/** + * Reply with attributes + * + * Possible requests: + * getattr, setattr + * + * @param req request handle + * @param attr the attributes + * @param attr_timeout validity timeout (in seconds) for the attributes + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_attr(fuse_req_t req, const struct stat *attr, + double attr_timeout); + +/** + * Reply with the contents of a symbolic link + * + * Possible requests: + * readlink + * + * @param req request handle + * @param link symbolic link contents + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_readlink(fuse_req_t req, const char *link); + +/** + * Reply with open parameters + * + * currently the following members of 'fi' are used: + * fh, direct_io, keep_cache + * + * Possible requests: + * open, opendir + * + * @param req request handle + * @param fi file information + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_open(fuse_req_t req, const struct fuse_file_info *fi); + +/** + * Reply with number of bytes written + * + * Possible requests: + * write + * + * @param req request handle + * @param count the number of bytes written + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_write(fuse_req_t req, size_t count); + +/** + * Reply with data + * + * Possible requests: + * read, readdir, getxattr, listxattr + * + * @param req request handle + * @param buf buffer containing data + * @param size the size of data in bytes + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_buf(fuse_req_t req, const char *buf, size_t size); + +/** + * Reply with data copied/moved from buffer(s) + * + * Zero copy data transfer ("splicing") will be used under + * the following circumstances: + * + * 1. FUSE_CAP_SPLICE_WRITE is set in fuse_conn_info.want, and + * 2. the kernel supports splicing from the fuse device + * (FUSE_CAP_SPLICE_WRITE is set in fuse_conn_info.capable), and + * 3. *flags* does not contain FUSE_BUF_NO_SPLICE + * 4. The amount of data that is provided in file-descriptor backed + * buffers (i.e., buffers for which bufv[n].flags == FUSE_BUF_FD) + * is at least twice the page size. + * + * In order for SPLICE_F_MOVE to be used, the following additional + * conditions have to be fulfilled: + * + * 1. FUSE_CAP_SPLICE_MOVE is set in fuse_conn_info.want, and + * 2. the kernel supports it (i.e, FUSE_CAP_SPLICE_MOVE is set in + fuse_conn_info.capable), and + * 3. *flags* contains FUSE_BUF_SPLICE_MOVE + * + * Note that, if splice is used, the data is actually spliced twice: + * once into a temporary pipe (to prepend header data), and then again + * into the kernel. If some of the provided buffers are memory-backed, + * the data in them is copied in step one and spliced in step two. + * + * The FUSE_BUF_SPLICE_FORCE_SPLICE and FUSE_BUF_SPLICE_NONBLOCK flags + * are silently ignored. + * + * Possible requests: + * read, readdir, getxattr, listxattr + * + * Side effects: + * when used to return data from a readdirplus() (but not readdir()) + * call, increments the lookup count of each returned entry by one + * on success. + * + * @param req request handle + * @param bufv buffer vector + * @param flags flags controlling the copy + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_data(fuse_req_t req, struct fuse_bufvec *bufv, + enum fuse_buf_copy_flags flags); + +/** + * Reply with data vector + * + * Possible requests: + * read, readdir, getxattr, listxattr + * + * @param req request handle + * @param iov the vector containing the data + * @param count the size of vector + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_iov(fuse_req_t req, const struct iovec *iov, int count); + +/** + * Reply with filesystem statistics + * + * Possible requests: + * statfs + * + * @param req request handle + * @param stbuf filesystem statistics + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_statfs(fuse_req_t req, const struct statvfs *stbuf); + +/** + * Reply with needed buffer size + * + * Possible requests: + * getxattr, listxattr + * + * @param req request handle + * @param count the buffer size needed in bytes + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_xattr(fuse_req_t req, size_t count); + +/** + * Reply with file lock information + * + * Possible requests: + * getlk + * + * @param req request handle + * @param lock the lock information + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_lock(fuse_req_t req, const struct flock *lock); + +/** + * Reply with block index + * + * Possible requests: + * bmap + * + * @param req request handle + * @param idx block index within device + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_bmap(fuse_req_t req, uint64_t idx); + +/* ----------------------------------------------------------- * + * Filling a buffer in readdir * + * ----------------------------------------------------------- */ + +/** + * Add a directory entry to the buffer + * + * Buffer needs to be large enough to hold the entry. If it's not, + * then the entry is not filled in but the size of the entry is still + * returned. The caller can check this by comparing the bufsize + * parameter with the returned entry size. If the entry size is + * larger than the buffer size, the operation failed. + * + * From the 'stbuf' argument the st_ino field and bits 12-15 of the + * st_mode field are used. The other fields are ignored. + * + * *off* should be any non-zero value that the filesystem can use to + * identify the current point in the directory stream. It does not + * need to be the actual physical position. A value of zero is + * reserved to mean "from the beginning", and should therefore never + * be used (the first call to fuse_add_direntry should be passed the + * offset of the second directory entry). + * + * @param req request handle + * @param buf the point where the new entry will be added to the buffer + * @param bufsize remaining size of the buffer + * @param name the name of the entry + * @param stbuf the file attributes + * @param off the offset of the next entry + * @return the space needed for the entry + */ +size_t fuse_add_direntry(fuse_req_t req, char *buf, size_t bufsize, + const char *name, const struct stat *stbuf, + off_t off); + +/** + * Add a directory entry to the buffer with the attributes + * + * See documentation of `fuse_add_direntry()` for more details. + * + * @param req request handle + * @param buf the point where the new entry will be added to the buffer + * @param bufsize remaining size of the buffer + * @param name the name of the entry + * @param e the directory entry + * @param off the offset of the next entry + * @return the space needed for the entry + */ +size_t fuse_add_direntry_plus(fuse_req_t req, char *buf, size_t bufsize, + const char *name, + const struct fuse_entry_param *e, off_t off); + +/** + * Reply to ask for data fetch and output buffer preparation. ioctl + * will be retried with the specified input data fetched and output + * buffer prepared. + * + * Possible requests: + * ioctl + * + * @param req request handle + * @param in_iov iovec specifying data to fetch from the caller + * @param in_count number of entries in in_iov + * @param out_iov iovec specifying addresses to write output to + * @param out_count number of entries in out_iov + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_ioctl_retry(fuse_req_t req, + const struct iovec *in_iov, size_t in_count, + const struct iovec *out_iov, size_t out_count); + +/** + * Reply to finish ioctl + * + * Possible requests: + * ioctl + * + * @param req request handle + * @param result result to be passed to the caller + * @param buf buffer containing output data + * @param size length of output data + */ +int fuse_reply_ioctl(fuse_req_t req, int result, const void *buf, size_t size); + +/** + * Reply to finish ioctl with iov buffer + * + * Possible requests: + * ioctl + * + * @param req request handle + * @param result result to be passed to the caller + * @param iov the vector containing the data + * @param count the size of vector + */ +int fuse_reply_ioctl_iov(fuse_req_t req, int result, const struct iovec *iov, + int count); + +/** + * Reply with poll result event mask + * + * @param req request handle + * @param revents poll result event mask + */ +int fuse_reply_poll(fuse_req_t req, unsigned revents); + +/** + * Reply with offset + * + * Possible requests: + * lseek + * + * @param req request handle + * @param off offset of next data or hole + * @return zero for success, -errno for failure to send reply + */ +int fuse_reply_lseek(fuse_req_t req, off_t off); + +/* ----------------------------------------------------------- * + * Notification * + * ----------------------------------------------------------- */ + +/** + * Notify IO readiness event + * + * For more information, please read comment for poll operation. + * + * @param ph poll handle to notify IO readiness event for + */ +int fuse_lowlevel_notify_poll(struct fuse_pollhandle *ph); + +/** + * Notify to invalidate cache for an inode. + * + * Added in FUSE protocol version 7.12. If the kernel does not support + * this (or a newer) version, the function will return -ENOSYS and do + * nothing. + * + * If the filesystem has writeback caching enabled, invalidating an + * inode will first trigger a writeback of all dirty pages. The call + * will block until all writeback requests have completed and the + * inode has been invalidated. It will, however, not wait for + * completion of pending writeback requests that have been issued + * before. + * + * If there are no dirty pages, this function will never block. + * + * @param se the session object + * @param ino the inode number + * @param off the offset in the inode where to start invalidating + * or negative to invalidate attributes only + * @param len the amount of cache to invalidate or 0 for all + * @return zero for success, -errno for failure + */ +int fuse_lowlevel_notify_inval_inode(struct fuse_session *se, fuse_ino_t ino, + off_t off, off_t len); + +/** + * Notify to invalidate parent attributes and the dentry matching + * parent/name + * + * To avoid a deadlock this function must not be called in the + * execution path of a related filesytem operation or within any code + * that could hold a lock that could be needed to execute such an + * operation. As of kernel 4.18, a "related operation" is a lookup(), + * symlink(), mknod(), mkdir(), unlink(), rename(), link() or create() + * request for the parent, and a setattr(), unlink(), rmdir(), + * rename(), setxattr(), removexattr(), readdir() or readdirplus() + * request for the inode itself. + * + * When called correctly, this function will never block. + * + * Added in FUSE protocol version 7.12. If the kernel does not support + * this (or a newer) version, the function will return -ENOSYS and do + * nothing. + * + * @param se the session object + * @param parent inode number + * @param name file name + * @param namelen strlen() of file name + * @return zero for success, -errno for failure + */ +int fuse_lowlevel_notify_inval_entry(struct fuse_session *se, fuse_ino_t parent, + const char *name, size_t namelen); + +/** + * This function behaves like fuse_lowlevel_notify_inval_entry() with + * the following additional effect (at least as of Linux kernel 4.8): + * + * If the provided *child* inode matches the inode that is currently + * associated with the cached dentry, and if there are any inotify + * watches registered for the dentry, then the watchers are informed + * that the dentry has been deleted. + * + * To avoid a deadlock this function must not be called while + * executing a related filesytem operation or while holding a lock + * that could be needed to execute such an operation (see the + * description of fuse_lowlevel_notify_inval_entry() for more + * details). + * + * When called correctly, this function will never block. + * + * Added in FUSE protocol version 7.18. If the kernel does not support + * this (or a newer) version, the function will return -ENOSYS and do + * nothing. + * + * @param se the session object + * @param parent inode number + * @param child inode number + * @param name file name + * @param namelen strlen() of file name + * @return zero for success, -errno for failure + */ +int fuse_lowlevel_notify_delete(struct fuse_session *se, + fuse_ino_t parent, fuse_ino_t child, + const char *name, size_t namelen); + +/** + * Store data to the kernel buffers + * + * Synchronously store data in the kernel buffers belonging to the + * given inode. The stored data is marked up-to-date (no read will be + * performed against it, unless it's invalidated or evicted from the + * cache). + * + * If the stored data overflows the current file size, then the size + * is extended, similarly to a write(2) on the filesystem. + * + * If this function returns an error, then the store wasn't fully + * completed, but it may have been partially completed. + * + * Added in FUSE protocol version 7.15. If the kernel does not support + * this (or a newer) version, the function will return -ENOSYS and do + * nothing. + * + * @param se the session object + * @param ino the inode number + * @param offset the starting offset into the file to store to + * @param bufv buffer vector + * @param flags flags controlling the copy + * @return zero for success, -errno for failure + */ +int fuse_lowlevel_notify_store(struct fuse_session *se, fuse_ino_t ino, + off_t offset, struct fuse_bufvec *bufv, + enum fuse_buf_copy_flags flags); +/** + * Retrieve data from the kernel buffers + * + * Retrieve data in the kernel buffers belonging to the given inode. + * If successful then the retrieve_reply() method will be called with + * the returned data. + * + * Only present pages are returned in the retrieve reply. Retrieving + * stops when it finds a non-present page and only data prior to that + * is returned. + * + * If this function returns an error, then the retrieve will not be + * completed and no reply will be sent. + * + * This function doesn't change the dirty state of pages in the kernel + * buffer. For dirty pages the write() method will be called + * regardless of having been retrieved previously. + * + * Added in FUSE protocol version 7.15. If the kernel does not support + * this (or a newer) version, the function will return -ENOSYS and do + * nothing. + * + * @param se the session object + * @param ino the inode number + * @param size the number of bytes to retrieve + * @param offset the starting offset into the file to retrieve from + * @param cookie user data to supply to the reply callback + * @return zero for success, -errno for failure + */ +int fuse_lowlevel_notify_retrieve(struct fuse_session *se, fuse_ino_t ino, + size_t size, off_t offset, void *cookie); + + +/* ----------------------------------------------------------- * + * Utility functions * + * ----------------------------------------------------------- */ + +/** + * Get the userdata from the request + * + * @param req request handle + * @return the user data passed to fuse_session_new() + */ +void *fuse_req_userdata(fuse_req_t req); + +/** + * Get the context from the request + * + * The pointer returned by this function will only be valid for the + * request's lifetime + * + * @param req request handle + * @return the context structure + */ +const struct fuse_ctx *fuse_req_ctx(fuse_req_t req); + +/** + * Get the current supplementary group IDs for the specified request + * + * Similar to the getgroups(2) system call, except the return value is + * always the total number of group IDs, even if it is larger than the + * specified size. + * + * The current fuse kernel module in linux (as of 2.6.30) doesn't pass + * the group list to userspace, hence this function needs to parse + * "/proc/$TID/task/$TID/status" to get the group IDs. + * + * This feature may not be supported on all operating systems. In + * such a case this function will return -ENOSYS. + * + * @param req request handle + * @param size size of given array + * @param list array of group IDs to be filled in + * @return the total number of supplementary group IDs or -errno on failure + */ +int fuse_req_getgroups(fuse_req_t req, int size, gid_t list[]); + +/** + * Callback function for an interrupt + * + * @param req interrupted request + * @param data user data + */ +typedef void (*fuse_interrupt_func_t)(fuse_req_t req, void *data); + +/** + * Register/unregister callback for an interrupt + * + * If an interrupt has already happened, then the callback function is + * called from within this function, hence it's not possible for + * interrupts to be lost. + * + * @param req request handle + * @param func the callback function or NULL for unregister + * @param data user data passed to the callback function + */ +void fuse_req_interrupt_func(fuse_req_t req, fuse_interrupt_func_t func, + void *data); + +/** + * Check if a request has already been interrupted + * + * @param req request handle + * @return 1 if the request has been interrupted, 0 otherwise + */ +int fuse_req_interrupted(fuse_req_t req); + + +/* ----------------------------------------------------------- * + * Inquiry functions * + * ----------------------------------------------------------- */ + +/** + * Print low-level version information to stdout. + */ +void fuse_lowlevel_version(void); + +/** + * Print available low-level options to stdout. This is not an + * exhaustive list, but includes only those options that may be of + * interest to an end-user of a file system. + */ +void fuse_lowlevel_help(void); + +/** + * Print available options for `fuse_parse_cmdline()`. + */ +void fuse_cmdline_help(void); + +/* ----------------------------------------------------------- * + * Filesystem setup & teardown * + * ----------------------------------------------------------- */ + +struct fuse_cmdline_opts { + int singlethread; + int foreground; + int debug; + int nodefault_subtype; + char *mountpoint; + int show_version; + int show_help; + int clone_fd; + unsigned int max_idle_threads; +}; + +/** + * Utility function to parse common options for simple file systems + * using the low-level API. A help text that describes the available + * options can be printed with `fuse_cmdline_help`. A single + * non-option argument is treated as the mountpoint. Multiple + * non-option arguments will result in an error. + * + * If neither -o subtype= or -o fsname= options are given, a new + * subtype option will be added and set to the basename of the program + * (the fsname will remain unset, and then defaults to "fuse"). + * + * Known options will be removed from *args*, unknown options will + * remain. + * + * @param args argument vector (input+output) + * @param opts output argument for parsed options + * @return 0 on success, -1 on failure + */ +int fuse_parse_cmdline(struct fuse_args *args, + struct fuse_cmdline_opts *opts); + +/** + * Create a low level session. + * + * Returns a session structure suitable for passing to + * fuse_session_mount() and fuse_session_loop(). + * + * This function accepts most file-system independent mount options + * (like context, nodev, ro - see mount(8)), as well as the general + * fuse mount options listed in mount.fuse(8) (e.g. -o allow_root and + * -o default_permissions, but not ``-o use_ino``). Instead of `-o + * debug`, debugging may also enabled with `-d` or `--debug`. + * + * If not all options are known, an error message is written to stderr + * and the function returns NULL. + * + * Option parsing skips argv[0], which is assumed to contain the + * program name. To prevent accidentally passing an option in + * argv[0], this element must always be present (even if no options + * are specified). It may be set to the empty string ('\0') if no + * reasonable value can be provided. + * + * @param args argument vector + * @param op the (low-level) filesystem operations + * @param op_size sizeof(struct fuse_lowlevel_ops) + * @param userdata user data + * + * @return the fuse session on success, NULL on failure + **/ +struct fuse_session *fuse_session_new(struct fuse_args *args, + const struct fuse_lowlevel_ops *op, + size_t op_size, void *userdata); + +/** + * Mount a FUSE file system. + * + * @param mountpoint the mount point path + * @param se session object + * + * @return 0 on success, -1 on failure. + **/ +int fuse_session_mount(struct fuse_session *se, const char *mountpoint); + +/** + * Enter a single threaded, blocking event loop. + * + * When the event loop terminates because the connection to the FUSE + * kernel module has been closed, this function returns zero. This + * happens when the filesystem is unmounted regularly (by the + * filesystem owner or root running the umount(8) or fusermount(1) + * command), or if connection is explicitly severed by writing ``1`` + * to the``abort`` file in ``/sys/fs/fuse/connections/NNN``. The only + * way to distinguish between these two conditions is to check if the + * filesystem is still mounted after the session loop returns. + * + * When some error occurs during request processing, the function + * returns a negated errno(3) value. + * + * If the loop has been terminated because of a signal handler + * installed by fuse_set_signal_handlers(), this function returns the + * (positive) signal value that triggered the exit. + * + * @param se the session + * @return 0, -errno, or a signal value + */ +int fuse_session_loop(struct fuse_session *se); + +/** + * Enter a multi-threaded event loop. + * + * For a description of the return value and the conditions when the + * event loop exits, refer to the documentation of + * fuse_session_loop(). + * + * @param se the session + * @param config session loop configuration + * @return see fuse_session_loop() + */ +#if FUSE_USE_VERSION < 32 +int fuse_session_loop_mt_31(struct fuse_session *se, int clone_fd); +#define fuse_session_loop_mt(se, clone_fd) fuse_session_loop_mt_31(se, clone_fd) +#else +int fuse_session_loop_mt(struct fuse_session *se, struct fuse_loop_config *config); +#endif + +/** + * Flag a session as terminated. + * + * This function is invoked by the POSIX signal handlers, when + * registered using fuse_set_signal_handlers(). It will cause any + * running event loops to terminate on the next opportunity. + * + * @param se the session + */ +void fuse_session_exit(struct fuse_session *se); + +/** + * Reset the terminated flag of a session + * + * @param se the session + */ +void fuse_session_reset(struct fuse_session *se); + +/** + * Query the terminated flag of a session + * + * @param se the session + * @return 1 if exited, 0 if not exited + */ +int fuse_session_exited(struct fuse_session *se); + +/** + * Ensure that file system is unmounted. + * + * In regular operation, the file system is typically unmounted by the + * user calling umount(8) or fusermount(1), which then terminates the + * FUSE session loop. However, the session loop may also terminate as + * a result of an explicit call to fuse_session_exit() (e.g. by a + * signal handler installed by fuse_set_signal_handler()). In this + * case the filesystem remains mounted, but any attempt to access it + * will block (while the filesystem process is still running) or give + * an ESHUTDOWN error (after the filesystem process has terminated). + * + * If the communication channel with the FUSE kernel module is still + * open (i.e., if the session loop was terminated by an explicit call + * to fuse_session_exit()), this function will close it and unmount + * the filesystem. If the communication channel has been closed by the + * kernel, this method will do (almost) nothing. + * + * NOTE: The above semantics mean that if the connection to the kernel + * is terminated via the ``/sys/fs/fuse/connections/NNN/abort`` file, + * this method will *not* unmount the filesystem. + * + * @param se the session + */ +void fuse_session_unmount(struct fuse_session *se); + +/** + * Destroy a session + * + * @param se the session + */ +void fuse_session_destroy(struct fuse_session *se); + +/* ----------------------------------------------------------- * + * Custom event loop support * + * ----------------------------------------------------------- */ + +/** + * Return file descriptor for communication with kernel. + * + * The file selector can be used to integrate FUSE with a custom event + * loop. Whenever data is available for reading on the provided fd, + * the event loop should call `fuse_session_receive_buf` followed by + * `fuse_session_process_buf` to process the request. + * + * The returned file descriptor is valid until `fuse_session_unmount` + * is called. + * + * @param se the session + * @return a file descriptor + */ +int fuse_session_fd(struct fuse_session *se); + +/** + * Process a raw request supplied in a generic buffer + * + * The fuse_buf may contain a memory buffer or a pipe file descriptor. + * + * @param se the session + * @param buf the fuse_buf containing the request + */ +void fuse_session_process_buf(struct fuse_session *se, + const struct fuse_buf *buf); + +/** + * Read a raw request from the kernel into the supplied buffer. + * + * Depending on file system options, system capabilities, and request + * size the request is either read into a memory buffer or spliced + * into a temporary pipe. + * + * @param se the session + * @param buf the fuse_buf to store the request in + * @return the actual size of the raw request, or -errno on error + */ +int fuse_session_receive_buf(struct fuse_session *se, struct fuse_buf *buf); + +#ifdef __cplusplus +} +#endif + +#endif /* FUSE_LOWLEVEL_H_ */ diff --git a/tools/virtiofsd/fuse_misc.h b/tools/virtiofsd/fuse_misc.h new file mode 100644 index 0000000..2f6663e --- /dev/null +++ b/tools/virtiofsd/fuse_misc.h @@ -0,0 +1,59 @@ +/* + FUSE: Filesystem in Userspace + Copyright (C) 2001-2007 Miklos Szeredi + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB +*/ + +#include + +/* + Versioned symbols cannot be used in some cases because it + - confuse the dynamic linker in uClibc + - not supported on MacOSX (in MachO binary format) +*/ +#if (!defined(__UCLIBC__) && !defined(__APPLE__)) +#define FUSE_SYMVER(x) __asm__(x) +#else +#define FUSE_SYMVER(x) +#endif + +#ifndef USE_UCLIBC +#define fuse_mutex_init(mut) pthread_mutex_init(mut, NULL) +#else +/* Is this hack still needed? */ +static inline void fuse_mutex_init(pthread_mutex_t *mut) +{ + pthread_mutexattr_t attr; + pthread_mutexattr_init(&attr); + pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_ADAPTIVE_NP); + pthread_mutex_init(mut, &attr); + pthread_mutexattr_destroy(&attr); +} +#endif + +#ifdef HAVE_STRUCT_STAT_ST_ATIM +/* Linux */ +#define ST_ATIM_NSEC(stbuf) ((stbuf)->st_atim.tv_nsec) +#define ST_CTIM_NSEC(stbuf) ((stbuf)->st_ctim.tv_nsec) +#define ST_MTIM_NSEC(stbuf) ((stbuf)->st_mtim.tv_nsec) +#define ST_ATIM_NSEC_SET(stbuf, val) (stbuf)->st_atim.tv_nsec = (val) +#define ST_CTIM_NSEC_SET(stbuf, val) (stbuf)->st_ctim.tv_nsec = (val) +#define ST_MTIM_NSEC_SET(stbuf, val) (stbuf)->st_mtim.tv_nsec = (val) +#elif defined(HAVE_STRUCT_STAT_ST_ATIMESPEC) +/* FreeBSD */ +#define ST_ATIM_NSEC(stbuf) ((stbuf)->st_atimespec.tv_nsec) +#define ST_CTIM_NSEC(stbuf) ((stbuf)->st_ctimespec.tv_nsec) +#define ST_MTIM_NSEC(stbuf) ((stbuf)->st_mtimespec.tv_nsec) +#define ST_ATIM_NSEC_SET(stbuf, val) (stbuf)->st_atimespec.tv_nsec = (val) +#define ST_CTIM_NSEC_SET(stbuf, val) (stbuf)->st_ctimespec.tv_nsec = (val) +#define ST_MTIM_NSEC_SET(stbuf, val) (stbuf)->st_mtimespec.tv_nsec = (val) +#else +#define ST_ATIM_NSEC(stbuf) 0 +#define ST_CTIM_NSEC(stbuf) 0 +#define ST_MTIM_NSEC(stbuf) 0 +#define ST_ATIM_NSEC_SET(stbuf, val) do { } while (0) +#define ST_CTIM_NSEC_SET(stbuf, val) do { } while (0) +#define ST_MTIM_NSEC_SET(stbuf, val) do { } while (0) +#endif diff --git a/tools/virtiofsd/fuse_opt.h b/tools/virtiofsd/fuse_opt.h new file mode 100644 index 0000000..d8573e7 --- /dev/null +++ b/tools/virtiofsd/fuse_opt.h @@ -0,0 +1,271 @@ +/* + FUSE: Filesystem in Userspace + Copyright (C) 2001-2007 Miklos Szeredi + + This program can be distributed under the terms of the GNU LGPLv2. + See the file COPYING.LIB. +*/ + +#ifndef FUSE_OPT_H_ +#define FUSE_OPT_H_ + +/** @file + * + * This file defines the option parsing interface of FUSE + */ + +#ifdef __cplusplus +extern "C" { +#endif + +/** + * Option description + * + * This structure describes a single option, and action associated + * with it, in case it matches. + * + * More than one such match may occur, in which case the action for + * each match is executed. + * + * There are three possible actions in case of a match: + * + * i) An integer (int or unsigned) variable determined by 'offset' is + * set to 'value' + * + * ii) The processing function is called, with 'value' as the key + * + * iii) An integer (any) or string (char *) variable determined by + * 'offset' is set to the value of an option parameter + * + * 'offset' should normally be either set to + * + * - 'offsetof(struct foo, member)' actions i) and iii) + * + * - -1 action ii) + * + * The 'offsetof()' macro is defined in the header. + * + * The template determines which options match, and also have an + * effect on the action. Normally the action is either i) or ii), but + * if a format is present in the template, then action iii) is + * performed. + * + * The types of templates are: + * + * 1) "-x", "-foo", "--foo", "--foo-bar", etc. These match only + * themselves. Invalid values are "--" and anything beginning + * with "-o" + * + * 2) "foo", "foo-bar", etc. These match "-ofoo", "-ofoo-bar" or + * the relevant option in a comma separated option list + * + * 3) "bar=", "--foo=", etc. These are variations of 1) and 2) + * which have a parameter + * + * 4) "bar=%s", "--foo=%lu", etc. Same matching as above but perform + * action iii). + * + * 5) "-x ", etc. Matches either "-xparam" or "-x param" as + * two separate arguments + * + * 6) "-x %s", etc. Combination of 4) and 5) + * + * If the format is "%s", memory is allocated for the string unlike with + * scanf(). The previous value (if non-NULL) stored at the this location is + * freed. + */ +struct fuse_opt { + /** Matching template and optional parameter formatting */ + const char *templ; + + /** + * Offset of variable within 'data' parameter of fuse_opt_parse() + * or -1 + */ + unsigned long offset; + + /** + * Value to set the variable to, or to be passed as 'key' to the + * processing function. Ignored if template has a format + */ + int value; +}; + +/** + * Key option. In case of a match, the processing function will be + * called with the specified key. + */ +#define FUSE_OPT_KEY(templ, key) { templ, -1U, key } + +/** + * Last option. An array of 'struct fuse_opt' must end with a NULL + * template value + */ +#define FUSE_OPT_END { NULL, 0, 0 } + +/** + * Argument list + */ +struct fuse_args { + /** Argument count */ + int argc; + + /** Argument vector. NULL terminated */ + char **argv; + + /** Is 'argv' allocated? */ + int allocated; +}; + +/** + * Initializer for 'struct fuse_args' + */ +#define FUSE_ARGS_INIT(argc, argv) { argc, argv, 0 } + +/** + * Key value passed to the processing function if an option did not + * match any template + */ +#define FUSE_OPT_KEY_OPT -1 + +/** + * Key value passed to the processing function for all non-options + * + * Non-options are the arguments beginning with a character other than + * '-' or all arguments after the special '--' option + */ +#define FUSE_OPT_KEY_NONOPT -2 + +/** + * Special key value for options to keep + * + * Argument is not passed to processing function, but behave as if the + * processing function returned 1 + */ +#define FUSE_OPT_KEY_KEEP -3 + +/** + * Special key value for options to discard + * + * Argument is not passed to processing function, but behave as if the + * processing function returned zero + */ +#define FUSE_OPT_KEY_DISCARD -4 + +/** + * Processing function + * + * This function is called if + * - option did not match any 'struct fuse_opt' + * - argument is a non-option + * - option did match and offset was set to -1 + * + * The 'arg' parameter will always contain the whole argument or + * option including the parameter if exists. A two-argument option + * ("-x foo") is always converted to single argument option of the + * form "-xfoo" before this function is called. + * + * Options of the form '-ofoo' are passed to this function without the + * '-o' prefix. + * + * The return value of this function determines whether this argument + * is to be inserted into the output argument vector, or discarded. + * + * @param data is the user data passed to the fuse_opt_parse() function + * @param arg is the whole argument or option + * @param key determines why the processing function was called + * @param outargs the current output argument list + * @return -1 on error, 0 if arg is to be discarded, 1 if arg should be kept + */ +typedef int (*fuse_opt_proc_t)(void *data, const char *arg, int key, + struct fuse_args *outargs); + +/** + * Option parsing function + * + * If 'args' was returned from a previous call to fuse_opt_parse() or + * it was constructed from + * + * A NULL 'args' is equivalent to an empty argument vector + * + * A NULL 'opts' is equivalent to an 'opts' array containing a single + * end marker + * + * A NULL 'proc' is equivalent to a processing function always + * returning '1' + * + * @param args is the input and output argument list + * @param data is the user data + * @param opts is the option description array + * @param proc is the processing function + * @return -1 on error, 0 on success + */ +int fuse_opt_parse(struct fuse_args *args, void *data, + const struct fuse_opt opts[], fuse_opt_proc_t proc); + +/** + * Add an option to a comma separated option list + * + * @param opts is a pointer to an option list, may point to a NULL value + * @param opt is the option to add + * @return -1 on allocation error, 0 on success + */ +int fuse_opt_add_opt(char **opts, const char *opt); + +/** + * Add an option, escaping commas, to a comma separated option list + * + * @param opts is a pointer to an option list, may point to a NULL value + * @param opt is the option to add + * @return -1 on allocation error, 0 on success + */ +int fuse_opt_add_opt_escaped(char **opts, const char *opt); + +/** + * Add an argument to a NULL terminated argument vector + * + * @param args is the structure containing the current argument list + * @param arg is the new argument to add + * @return -1 on allocation error, 0 on success + */ +int fuse_opt_add_arg(struct fuse_args *args, const char *arg); + +/** + * Add an argument at the specified position in a NULL terminated + * argument vector + * + * Adds the argument to the N-th position. This is useful for adding + * options at the beginning of the array which must not come after the + * special '--' option. + * + * @param args is the structure containing the current argument list + * @param pos is the position at which to add the argument + * @param arg is the new argument to add + * @return -1 on allocation error, 0 on success + */ +int fuse_opt_insert_arg(struct fuse_args *args, int pos, const char *arg); + +/** + * Free the contents of argument list + * + * The structure itself is not freed + * + * @param args is the structure containing the argument list + */ +void fuse_opt_free_args(struct fuse_args *args); + + +/** + * Check if an option matches + * + * @param opts is the option description array + * @param opt is the option to match + * @return 1 if a match is found, 0 if not + */ +int fuse_opt_match(const struct fuse_opt opts[], const char *opt); + +#ifdef __cplusplus +} +#endif + +#endif /* FUSE_OPT_H_ */ diff --git a/tools/virtiofsd/passthrough_helpers.h b/tools/virtiofsd/passthrough_helpers.h new file mode 100644 index 0000000..6b77c33 --- /dev/null +++ b/tools/virtiofsd/passthrough_helpers.h @@ -0,0 +1,76 @@ +/* + * FUSE: Filesystem in Userspace + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE + */ + +/* + * Creates files on the underlying file system in response to a FUSE_MKNOD + * operation + */ +static int mknod_wrapper(int dirfd, const char *path, const char *link, + int mode, dev_t rdev) +{ + int res; + + if (S_ISREG(mode)) { + res = openat(dirfd, path, O_CREAT | O_EXCL | O_WRONLY, mode); + if (res >= 0) + res = close(res); + } else if (S_ISDIR(mode)) { + res = mkdirat(dirfd, path, mode); + } else if (S_ISLNK(mode) && link != NULL) { + res = symlinkat(link, dirfd, path); + } else if (S_ISFIFO(mode)) { + res = mkfifoat(dirfd, path, mode); +#ifdef __FreeBSD__ + } else if (S_ISSOCK(mode)) { + struct sockaddr_un su; + int fd; + + if (strlen(path) >= sizeof(su.sun_path)) { + errno = ENAMETOOLONG; + return -1; + } + fd = socket(AF_UNIX, SOCK_STREAM, 0); + if (fd >= 0) { + /* + * We must bind the socket to the underlying file + * system to create the socket file, even though + * we'll never listen on this socket. + */ + su.sun_family = AF_UNIX; + strncpy(su.sun_path, path, sizeof(su.sun_path)); + res = bindat(dirfd, fd, (struct sockaddr*)&su, + sizeof(su)); + if (res == 0) + close(fd); + } else { + res = -1; + } +#endif + } else { + res = mknodat(dirfd, path, mode, rdev); + } + + return res; +} -- 1.8.3.1