mirror of
https://github.com/trapexit/mergerfs.git
synced 2026-10-26 19:01:56 -04:00
Reinstate the splice read/write/move paths removed in 84592a9f, now
gated behind runtime options with negotiated pipe sizing and fallback
logging.
- Transport: msgbuf-piped splice receive, fd-referenced splice reply
(vmsplice + splice), splice-move pipe-bounce, and restored
fuse_buf_copy/fuse_buf_flags plumbing for copy fallbacks
- Negotiation: new pipe-max-size option (auto-raised at INIT when
possible; the sysctl read/write is factored into src/pipe_max_size.cpp
so fuse_init.cpp and option_parser.cpp share one implementation);
fuse-msg-size clamped to what the pipe can hold; splice is disabled
at INIT with a warning if the sysctl can't fit even the minimum,
retracting the FUSE_CAP_SPLICE_* bits already advertised to the
kernel in that case
- Read replies and splice-move are gated at 128 KiB (below that the
syscalls cost more than the copy they avoid); every fallback path
degrades to the copy path and logs once per process via SysLog
- msgbuf pipe cache is process-global and mutex-guarded (mutex_lockguard)
so pipes can move between read and worker threads under
process-thread mode
- API simplified after review: collapsed the three per-axis options
(splice-read/write/move) into a single splice boolean (default on);
no-splice-read/write/move and the bare splice-read/write/move
spellings stay registered as accepted no-ops so existing mount
options don't fail the mount, but they do NOT alter the splice
knob - upgraders with no-splice-* in their config get splice
enabled and must set splice=false explicitly (see
deprecated_options.md)
- Fixed several correctness issues found in parallel review: a
dropped write_buf registration that silently corrupted writes, a
dirty-pipe reuse bug in msgbuf_free, an iov double-send in
fuse_reply_data's splice-move fallback, and pipe-capacity math that
assumed a 4096-byte page size
- Modernized the transport/negotiation code to this codebase's C++
style: RAII (unique_ptr/DEFER) instead of malloc/free and manual fd
close(), std::from_chars/to_chars instead of stoul/snprintf,
_ReplyPipe's pipe ops as methods instead of free functions closing
over a thread_local global, and vendored/libfuse/lib/buffer.cpp (a
near-verbatim port of upstream's buffer.c) brought in line with the
rest of the fork
- Documented splice and pipe-max-size options (mkdocs, FAQ,
deprecated-options list); added tools/bench-splice.sh
- Benchmarked on tmpfs: seq_write +21%, seq_read +52%, small_write
+5%, small_read +12%; validated with a 1h soak (~271 GiB
hash-verified I/O, zero integrity failures), a live-mount round-trip
of I/O above the splice threshold, and the existing unit/smoke
suites at their pre-existing baseline
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
430 lines
12 KiB
C
430 lines
12 KiB
C
#pragma once
|
|
|
|
#include "extern_c.h"
|
|
#include "fuse_conn_info.hpp"
|
|
#include "fuse_opt.h"
|
|
#include "fuse_timeouts.h"
|
|
|
|
#include <stdint.h>
|
|
#include <sys/types.h>
|
|
|
|
/* This interface uses 64 bit off_t */
|
|
#if _FILE_OFFSET_BITS != 64
|
|
#error Please add -D_FILE_OFFSET_BITS=64 to your compile flags!
|
|
#endif
|
|
|
|
#define FUSE_DEFAULT_MAX_PAGES_PER_REQ 32
|
|
#define FUSE_DEFAULT_MAX_MAX_PAGES 256
|
|
|
|
EXTERN_C_BEGIN
|
|
|
|
/**
|
|
* Information about open files
|
|
*
|
|
* Changed in version 2.5
|
|
*/
|
|
typedef struct fuse_file_info_t fuse_file_info_t;
|
|
struct fuse_file_info_t
|
|
{
|
|
/** Open flags. Available in open() and release() */
|
|
int flags;
|
|
|
|
/** In case of a write operation indicates if this was caused by a
|
|
writepage */
|
|
uint32_t writepage : 1;
|
|
|
|
/** Can be filled in by open, to use direct I/O on this file.
|
|
Introduced in version 2.4 */
|
|
uint32_t direct_io : 1;
|
|
|
|
/** Can be filled in by open, to indicate, that cached file data
|
|
need not be invalidated. Introduced in version 2.4 */
|
|
uint32_t keep_cache : 1;
|
|
|
|
/** Indicates a flush operation. Set in flush operation, also
|
|
maybe set in highlevel lock operation and lowlevel release
|
|
operation. Introduced in version 2.6 */
|
|
uint32_t flush : 1;
|
|
|
|
/** Can be filled in by open, to indicate that the file is not
|
|
seekable. Introduced in version 2.8 */
|
|
uint32_t 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(). Introduced in version
|
|
2.9 */
|
|
uint32_t flock_release : 1;
|
|
|
|
/* Requests the kernel to cache entries returned by readdir */
|
|
uint32_t cache_readdir : 1;
|
|
|
|
uint32_t auto_cache : 1;
|
|
|
|
uint32_t parallel_direct_writes:1;
|
|
|
|
uint32_t noflush:1;
|
|
uint32_t passthrough:1;
|
|
|
|
/** File handle. May be filled in by filesystem in open().
|
|
Available in all other file operations */
|
|
uint64_t fh;
|
|
|
|
int32_t backing_id;
|
|
|
|
/** Lock owner id. Available in locking operations and flush */
|
|
uint64_t lock_owner;
|
|
};
|
|
|
|
/**
|
|
* Capability bits for 'fuse_conn_info_t.capable' and 'fuse_conn_info_t.want'
|
|
*
|
|
* FUSE_CAP_ASYNC_READ: filesystem supports asynchronous read requests
|
|
* FUSE_CAP_POSIX_LOCKS: filesystem supports "remote" locking
|
|
* FUSE_CAP_ATOMIC_O_TRUNC: filesystem handles the O_TRUNC open flag
|
|
* FUSE_CAP_EXPORT_SUPPORT: filesystem handles lookups of "." and ".."
|
|
* FUSE_CAP_BIG_WRITES: filesystem can handle write size larger than 4kB
|
|
* FUSE_CAP_DONT_MASK: don't apply umask to file mode on create operations
|
|
* FUSE_CAP_SPLICE_WRITE: ability to use splice() to write to the fuse device
|
|
* FUSE_CAP_SPLICE_MOVE: ability to move data to the fuse device with splice()
|
|
* FUSE_CAP_SPLICE_READ: ability to use splice() to read from the fuse device
|
|
* FUSE_CAP_IOCTL_DIR: ioctl support on directories
|
|
* FUSE_CAP_CACHE_SYMLINKS: cache READLINK responses
|
|
*/
|
|
#define FUSE_CAP_ASYNC_READ (1ULL << 0)
|
|
#define FUSE_CAP_POSIX_LOCKS (1ULL << 1)
|
|
#define FUSE_CAP_ATOMIC_O_TRUNC (1ULL << 3)
|
|
#define FUSE_CAP_EXPORT_SUPPORT (1ULL << 4)
|
|
#define FUSE_CAP_BIG_WRITES (1ULL << 5)
|
|
#define FUSE_CAP_DONT_MASK (1ULL << 6)
|
|
/* Same values as the FUSE_SPLICE_WRITE/MOVE/READ wire
|
|
flags in fuse_kernel.h */
|
|
#define FUSE_CAP_SPLICE_WRITE (1ULL << 7)
|
|
#define FUSE_CAP_SPLICE_MOVE (1ULL << 8)
|
|
#define FUSE_CAP_SPLICE_READ (1ULL << 9)
|
|
#define FUSE_CAP_FLOCK_LOCKS (1ULL << 10)
|
|
#define FUSE_CAP_IOCTL_DIR (1ULL << 11)
|
|
#define FUSE_CAP_READDIR_PLUS (1ULL << 13)
|
|
#define FUSE_CAP_READDIR_PLUS_AUTO (1ULL << 14)
|
|
#define FUSE_CAP_ASYNC_DIO (1ULL << 15)
|
|
#define FUSE_CAP_WRITEBACK_CACHE (1ULL << 16)
|
|
#define FUSE_CAP_PARALLEL_DIROPS (1ULL << 18)
|
|
#define FUSE_CAP_POSIX_ACL (1ULL << 19)
|
|
#define FUSE_CAP_CACHE_SYMLINKS (1ULL << 20)
|
|
#define FUSE_CAP_MAX_PAGES (1ULL << 21)
|
|
#define FUSE_CAP_SETXATTR_EXT (1ULL << 22)
|
|
#define FUSE_CAP_DIRECT_IO_ALLOW_MMAP (1ULL << 23)
|
|
#define FUSE_CAP_CREATE_SUPP_GROUP (1ULL << 24)
|
|
#define FUSE_CAP_PASSTHROUGH (1ULL << 25)
|
|
#define FUSE_CAP_HANDLE_KILLPRIV (1ULL << 26)
|
|
#define FUSE_CAP_HANDLE_KILLPRIV_V2 (1ULL << 27)
|
|
#define FUSE_CAP_ALLOW_IDMAP (1ULL << 28)
|
|
|
|
/**
|
|
* 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
|
|
|
|
struct fuse_session;
|
|
struct fuse_chan;
|
|
struct fuse_pollhandle_t;
|
|
typedef struct fuse_pollhandle_t fuse_pollhandle_t;
|
|
|
|
/**
|
|
* Mount a FUSE filesystem.
|
|
*
|
|
* This function creates a FUSE mount and returns the /dev/fuse file descriptor.
|
|
* The file descriptor is stored in the session for internal use.
|
|
*
|
|
* @param mountpoint the mount point path
|
|
* @param args argument vector
|
|
* @return the /dev/fuse file descriptor on success, -1 on failure
|
|
*/
|
|
int fuse_mount(const char *mountpoint, struct fuse_args *args);
|
|
|
|
/**
|
|
* Umount a FUSE mountpoint
|
|
*
|
|
* @param mountpoint the mount point path
|
|
* @param fd the /dev/fuse file descriptor (or -1 if already closed)
|
|
*/
|
|
void fuse_unmount(const char *mountpoint, int fd);
|
|
|
|
/**
|
|
* Parse common options
|
|
*
|
|
* The following options are parsed:
|
|
*
|
|
* '-f' foreground
|
|
* '-d' '-odebug' foreground, but keep the debug option
|
|
* '-h' '--help' help
|
|
* '-ho' help without header
|
|
* '-ofsname=..' file system name, if not present, then set to the program
|
|
* name
|
|
*
|
|
* All parameters may be NULL
|
|
*
|
|
* @param args argument vector
|
|
* @param mountpoint the returned mountpoint, should be freed after use
|
|
* @param foreground set to 1 if one of the relevant options is present
|
|
* @return 0 on success, -1 on failure
|
|
*/
|
|
int fuse_parse_cmdline(struct fuse_args *args,
|
|
char **mountpoint,
|
|
int *foreground);
|
|
|
|
/**
|
|
* 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);
|
|
|
|
/**
|
|
* Destroy poll handle
|
|
*
|
|
* @param ph the poll handle
|
|
*/
|
|
void fuse_pollhandle_destroy(fuse_pollhandle_t *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;
|
|
};
|
|
|
|
/* Gather buffer declarations */
|
|
#define FUSE_BUFVEC_INIT(size__) \
|
|
/* cppcheck-suppress unassignedVariable */ \
|
|
(struct fuse_bufvec) { \
|
|
/* .count= */ 1, \
|
|
/* .idx = */ 0, \
|
|
/* .off = */ 0, \
|
|
{ /* .size = */ (size__), /* .flags = */ (enum fuse_buf_flags) 0, \
|
|
/* .mem = */ nullptr, /* .fd = */ -1, /* .pos = */ 0, } \
|
|
}
|
|
|
|
struct fuse_bufvec {
|
|
/**
|
|
* Number of buffers in the array
|
|
*/
|
|
size_t count;
|
|
|
|
/**
|
|
* Current buffer index
|
|
*/
|
|
size_t idx;
|
|
|
|
/**
|
|
* Current offset in the buffer
|
|
*/
|
|
size_t off;
|
|
|
|
/**
|
|
* Array of buffers
|
|
*/
|
|
struct fuse_buf buf[1];
|
|
};
|
|
|
|
/**
|
|
* Get the total size of data in a fuse buffer vector
|
|
*
|
|
* @param bufv buffer vector
|
|
* @return computed number of bytes
|
|
*/
|
|
size_t fuse_buf_size(const struct fuse_bufvec *bufv);
|
|
|
|
/**
|
|
* Copy data from one buffer vector to another
|
|
*
|
|
* This is mostly useful for a filesystem that uses
|
|
* buffers to refer to the data being written or read
|
|
*
|
|
* @param dstv destination buffer vector
|
|
* @param srcv source buffer vector
|
|
* @param flags flags passed to the buffer copy
|
|
* @return actual number of bytes copied or -errno on error
|
|
*/
|
|
ssize_t fuse_buf_copy(struct fuse_bufvec *dstv, struct fuse_bufvec *srcv,
|
|
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.
|
|
*
|
|
* @param se the session to exit
|
|
* @return 0 on success, -1 on failure
|
|
*/
|
|
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()
|
|
*/
|
|
void fuse_remove_signal_handlers(struct fuse_session *se);
|
|
|
|
EXTERN_C_END
|
|
|
|
/* C++-only convenience, deliberately declared OUTSIDE EXTERN_C_BEGIN/
|
|
EXTERN_C_END above: this is not part of the C-linkage libfuse-
|
|
compatible API, just a small typed factory used by the splice
|
|
read/write handlers (src/fuse_read_buf.cpp, src/fuse_write_buf.cpp,
|
|
vendored/libfuse/lib/fuse.cpp's fuse_lib_write) so they build a
|
|
single-entry, fd-referencing fuse_bufvec through one call instead of
|
|
each hand-assigning buf[0].flags/fd/pos individually. */
|
|
static
|
|
inline
|
|
struct fuse_bufvec
|
|
fuse_bufvec_fd(const int fd_,
|
|
const off_t pos_,
|
|
const size_t size_,
|
|
const fuse_buf_flags flags_)
|
|
{
|
|
struct fuse_bufvec bufv = FUSE_BUFVEC_INIT(size_);
|
|
|
|
bufv.buf[0].flags = flags_;
|
|
bufv.buf[0].fd = fd_;
|
|
bufv.buf[0].pos = pos_;
|
|
|
|
return bufv;
|
|
}
|