rclone/vfs/file.go
Nick Craig-Wood 28a8ebce5b vfs: fix rename of directory containing files to be uploaded
Before this change, if you renamed a directory containg files yet to
be uploaded then deleted the directory the files would still be
uploaded.

This fixes the problem by changing the directory path in all the file
objects in a directory when it is renamed. This wasn't necessary until
we introduced virtual files and directories which lived beyond the
directory flush mechanism.

Fixes #6809
2023-03-07 11:40:50 +00:00

813 lines
20 KiB
Go

package vfs
import (
"context"
"errors"
"fmt"
"os"
"path"
"sync"
"sync/atomic"
"time"
"github.com/rclone/rclone/fs"
"github.com/rclone/rclone/fs/log"
"github.com/rclone/rclone/fs/operations"
"github.com/rclone/rclone/vfs/vfscommon"
)
// The File object is tightly coupled to the Dir object. Since they
// both have locks there is plenty of potential for deadlocks. In
// order to mitigate this, we use the following conventions
//
// File may **only** call these methods from Dir with the File lock
// held.
//
// Dir.Fs
// Dir.VFS
//
// (As these are read only and do not need to take the Dir mutex.)
//
// File may **not** call any other Dir methods with the File lock
// held. This preserves total lock ordering and makes File subordinate
// to Dir as far as locking is concerned, preventing deadlocks.
//
// File may **not** read any members of Dir directly.
// File represents a file
type File struct {
inode uint64 // inode number - read only
size int64 // size of file - read and written with atomic int64 - must be 64 bit aligned
muRW sync.Mutex // synchronize RWFileHandle.openPending(), RWFileHandle.close() and File.Remove
mu sync.RWMutex // protects the following
d *Dir // parent directory
dPath string // path of parent directory. NB dir rename means all Files are flushed
o fs.Object // NB o may be nil if file is being written
leaf string // leaf name of the object
writers []Handle // writers for this file
pendingModTime time.Time // will be applied once o becomes available, i.e. after file was written
pendingRenameFun func(ctx context.Context) error // will be run/renamed after all writers close
sys atomic.Value // user defined info to be attached here
nwriters int32 // len(writers) which is read/updated with atomic
appendMode bool // file was opened with O_APPEND
}
// newFile creates a new File
//
// o may be nil
func newFile(d *Dir, dPath string, o fs.Object, leaf string) *File {
f := &File{
d: d,
dPath: dPath,
o: o,
leaf: leaf,
inode: newInode(),
}
if o != nil {
f.size = o.Size()
}
return f
}
// String converts it to printable
func (f *File) String() string {
if f == nil {
return "<nil *File>"
}
return f.Path()
}
// IsFile returns true for File - satisfies Node interface
func (f *File) IsFile() bool {
return true
}
// IsDir returns false for File - satisfies Node interface
func (f *File) IsDir() bool {
return false
}
// Mode bits of the file or directory - satisfies Node interface
func (f *File) Mode() (mode os.FileMode) {
f.mu.RLock()
defer f.mu.RUnlock()
mode = f.d.vfs.Opt.FilePerms
if f.appendMode {
mode |= os.ModeAppend
}
return mode
}
// Name (base) of the directory - satisfies Node interface
func (f *File) Name() (name string) {
f.mu.RLock()
defer f.mu.RUnlock()
return f.leaf
}
// _path returns the full path of the file
// use when lock is held
func (f *File) _path() string {
return path.Join(f.dPath, f.leaf)
}
// Path returns the full path of the file
func (f *File) Path() string {
f.mu.RLock()
dPath, leaf := f.dPath, f.leaf
f.mu.RUnlock()
return path.Join(dPath, leaf)
}
// Sys returns underlying data source (can be nil) - satisfies Node interface
func (f *File) Sys() interface{} {
return f.sys.Load()
}
// SetSys sets the underlying data source (can be nil) - satisfies Node interface
func (f *File) SetSys(x interface{}) {
f.sys.Store(x)
}
// Inode returns the inode number - satisfies Node interface
func (f *File) Inode() uint64 {
return f.inode
}
// Node returns the Node associated with this - satisfies Noder interface
func (f *File) Node() Node {
return f
}
// renameDir - call when parent directory has been renamed
func (f *File) renameDir(dPath string) {
f.mu.RLock()
f.dPath = dPath
f.mu.RUnlock()
}
// applyPendingRename runs a previously set rename operation if there are no
// more remaining writers. Call without lock held.
func (f *File) applyPendingRename() {
f.mu.RLock()
fun := f.pendingRenameFun
writing := f._writingInProgress()
f.mu.RUnlock()
if fun == nil || writing {
return
}
fs.Debugf(f.Path(), "Running delayed rename now")
if err := fun(context.TODO()); err != nil {
fs.Errorf(f.Path(), "delayed File.Rename error: %v", err)
}
}
// rename attempts to immediately rename a file if there are no open writers.
// Otherwise it will queue the rename operation on the remote until no writers
// remain.
func (f *File) rename(ctx context.Context, destDir *Dir, newName string) error {
f.mu.RLock()
d := f.d
oldPendingRenameFun := f.pendingRenameFun
f.mu.RUnlock()
if features := d.Fs().Features(); features.Move == nil && features.Copy == nil {
err := fmt.Errorf("Fs %q can't rename files (no server-side Move or Copy)", d.Fs())
fs.Errorf(f.Path(), "Dir.Rename error: %v", err)
return err
}
oldPath := f.Path()
// File.mu is unlocked here to call Dir.Path()
newPath := path.Join(destDir.Path(), newName)
renameCall := func(ctx context.Context) (err error) {
// chain rename calls if any
if oldPendingRenameFun != nil {
err := oldPendingRenameFun(ctx)
if err != nil {
return err
}
}
f.mu.RLock()
o := f.o
d := f.d
f.mu.RUnlock()
var newObject fs.Object
// if o is nil then are writing the file so no need to rename the object
if o != nil {
if o.Remote() == newPath {
return nil // no need to rename
}
// do the move of the remote object
dstOverwritten, _ := d.Fs().NewObject(ctx, newPath)
newObject, err = operations.Move(ctx, d.Fs(), dstOverwritten, newPath, o)
if err != nil {
fs.Errorf(f.Path(), "File.Rename error: %v", err)
return err
}
// newObject can be nil here for example if --dry-run
if newObject == nil {
err = errors.New("rename failed: nil object returned")
fs.Errorf(f.Path(), "File.Rename %v", err)
return err
}
}
// Rename in the cache
if d.vfs.cache != nil && d.vfs.cache.Exists(oldPath) {
if err := d.vfs.cache.Rename(oldPath, newPath, newObject); err != nil {
fs.Infof(f.Path(), "File.Rename failed in Cache: %v", err)
}
}
// Update the node with the new details
fs.Debugf(f.Path(), "Updating file with %v %p", newObject, f)
// f.rename(destDir, newObject)
f.mu.Lock()
if newObject != nil {
f.o = newObject
}
f.pendingRenameFun = nil
f.mu.Unlock()
return nil
}
// rename the file object
dPath := destDir.Path()
f.mu.Lock()
f.d = destDir
f.dPath = dPath
f.leaf = newName
writing := f._writingInProgress()
f.mu.Unlock()
// Delay the rename if not using RW caching. For the minimal case we
// need to look in the cache to see if caching is in use.
CacheMode := d.vfs.Opt.CacheMode
if writing &&
(CacheMode < vfscommon.CacheModeMinimal ||
(CacheMode == vfscommon.CacheModeMinimal && !destDir.vfs.cache.Exists(oldPath))) {
fs.Debugf(oldPath, "File is currently open, delaying rename %p", f)
f.mu.Lock()
f.pendingRenameFun = renameCall
f.mu.Unlock()
return nil
}
return renameCall(ctx)
}
// addWriter adds a write handle to the file
func (f *File) addWriter(h Handle) {
f.mu.Lock()
f.writers = append(f.writers, h)
atomic.AddInt32(&f.nwriters, 1)
f.mu.Unlock()
}
// delWriter removes a write handle from the file
func (f *File) delWriter(h Handle) {
f.mu.Lock()
defer f.applyPendingRename()
defer f.mu.Unlock()
var found = -1
for i := range f.writers {
if f.writers[i] == h {
found = i
break
}
}
if found >= 0 {
f.writers = append(f.writers[:found], f.writers[found+1:]...)
atomic.AddInt32(&f.nwriters, -1)
} else {
fs.Debugf(f._path(), "File.delWriter couldn't find handle")
}
}
// activeWriters returns the number of writers on the file
//
// Note that we don't take the mutex here. If we do then we can get a
// deadlock.
func (f *File) activeWriters() int {
return int(atomic.LoadInt32(&f.nwriters))
}
// _roundModTime rounds the time passed in to the Precision of the
// underlying Fs
//
// It should be called with the lock held
func (f *File) _roundModTime(modTime time.Time) time.Time {
precision := f.d.f.Precision()
if precision == fs.ModTimeNotSupported {
return modTime
}
return modTime.Truncate(precision)
}
// ModTime returns the modified time of the file
//
// if NoModTime is set then it returns the mod time of the directory
func (f *File) ModTime() (modTime time.Time) {
f.mu.RLock()
d, o, pendingModTime := f.d, f.o, f.pendingModTime
f.mu.RUnlock()
if d.vfs.Opt.NoModTime {
return d.ModTime()
}
// Read the modtime from a dirty item if it exists
if f.d.vfs.Opt.CacheMode >= vfscommon.CacheModeMinimal {
if item := f.d.vfs.cache.DirtyItem(f._path()); item != nil {
modTime, err := item.GetModTime()
if err != nil {
fs.Errorf(f._path(), "ModTime: Item GetModTime failed: %v", err)
} else {
return f._roundModTime(modTime)
}
}
}
if !pendingModTime.IsZero() {
return f._roundModTime(pendingModTime)
}
if o == nil {
return time.Now()
}
return o.ModTime(context.TODO())
}
// nonNegative returns 0 if i is -ve, i otherwise
func nonNegative(i int64) int64 {
if i >= 0 {
return i
}
return 0
}
// Size of the file
func (f *File) Size() int64 {
f.mu.RLock()
defer f.mu.RUnlock()
// Read the size from a dirty item if it exists
if f.d.vfs.Opt.CacheMode >= vfscommon.CacheModeMinimal {
if item := f.d.vfs.cache.DirtyItem(f._path()); item != nil {
size, err := item.GetSize()
if err != nil {
fs.Errorf(f._path(), "Size: Item GetSize failed: %v", err)
} else {
return size
}
}
}
// if o is nil it isn't valid yet or there are writers, so return the size so far
if f._writingInProgress() {
return atomic.LoadInt64(&f.size)
}
return nonNegative(f.o.Size())
}
// SetModTime sets the modtime for the file
//
// if NoModTime is set then it does nothing
func (f *File) SetModTime(modTime time.Time) error {
f.mu.Lock()
defer f.mu.Unlock()
if f.d.vfs.Opt.NoModTime {
return nil
}
if f.d.vfs.Opt.ReadOnly {
return EROFS
}
f.pendingModTime = modTime
// set the time of the file in the cache
if f.d.vfs.cache != nil && f.d.vfs.cache.Exists(f._path()) {
f.d.vfs.cache.SetModTime(f._path(), f.pendingModTime)
}
// Only update the ModTime when there are no writers, setObject will do it
if !f._writingInProgress() {
return f._applyPendingModTime()
}
// queue up for later, hoping f.o becomes available
return nil
}
// Apply a pending mod time
// Call with the mutex held
func (f *File) _applyPendingModTime() error {
if f.pendingModTime.IsZero() {
return nil
}
defer func() { f.pendingModTime = time.Time{} }()
if f.o == nil {
return errors.New("cannot apply ModTime, file object is not available")
}
dt := f.pendingModTime.Sub(f.o.ModTime(context.Background()))
modifyWindow := f.o.Fs().Precision()
if dt < modifyWindow && dt > -modifyWindow {
fs.Debugf(f.o, "Not setting pending mod time %v as it is already set", f.pendingModTime)
return nil
}
// set the time of the object
err := f.o.SetModTime(context.TODO(), f.pendingModTime)
switch err {
case nil:
fs.Debugf(f.o, "Applied pending mod time %v OK", f.pendingModTime)
case fs.ErrorCantSetModTime, fs.ErrorCantSetModTimeWithoutDelete:
// do nothing, in order to not break "touch somefile" if it exists already
default:
fs.Errorf(f.o, "Failed to apply pending mod time %v: %v", f.pendingModTime, err)
return err
}
return nil
}
// Apply a pending mod time
func (f *File) applyPendingModTime() error {
f.mu.Lock()
defer f.mu.Unlock()
return f._applyPendingModTime()
}
// _writingInProgress returns true of there are any open writers
// Call with read lock held
func (f *File) _writingInProgress() bool {
return f.o == nil || len(f.writers) != 0
}
// writingInProgress returns true of there are any open writers
func (f *File) writingInProgress() bool {
f.mu.RLock()
defer f.mu.RUnlock()
return f.o == nil || len(f.writers) != 0
}
// Update the size while writing
func (f *File) setSize(n int64) {
atomic.StoreInt64(&f.size, n)
}
// Update the object when written and add it to the directory
func (f *File) setObject(o fs.Object) {
f.mu.Lock()
f.o = o
_ = f._applyPendingModTime()
d := f.d
f.mu.Unlock()
// Release File.mu before calling Dir method
d.addObject(f)
}
// Update the object but don't update the directory cache - for use by
// the directory cache
func (f *File) setObjectNoUpdate(o fs.Object) {
f.mu.Lock()
f.o = o
f.mu.Unlock()
}
// Get the current fs.Object - may be nil
func (f *File) getObject() fs.Object {
f.mu.RLock()
defer f.mu.RUnlock()
return f.o
}
// exists returns whether the file exists already
func (f *File) exists() bool {
f.mu.RLock()
defer f.mu.RUnlock()
return f.o != nil
}
// Wait for f.o to become non nil for a short time returning it or an
// error. Use when opening a read handle.
//
// Call without the mutex held
func (f *File) waitForValidObject() (o fs.Object, err error) {
for i := 0; i < 50; i++ {
f.mu.RLock()
o = f.o
nwriters := len(f.writers)
f.mu.RUnlock()
if o != nil {
return o, nil
}
if nwriters == 0 {
return nil, errors.New("can't open file - writer failed")
}
time.Sleep(100 * time.Millisecond)
}
return nil, ENOENT
}
// openRead open the file for read
func (f *File) openRead() (fh *ReadFileHandle, err error) {
// if o is nil it isn't valid yet
_, err = f.waitForValidObject()
if err != nil {
return nil, err
}
// fs.Debugf(f.Path(), "File.openRead")
fh, err = newReadFileHandle(f)
if err != nil {
fs.Debugf(f.Path(), "File.openRead failed: %v", err)
return nil, err
}
return fh, nil
}
// openWrite open the file for write
func (f *File) openWrite(flags int) (fh *WriteFileHandle, err error) {
f.mu.RLock()
d := f.d
f.mu.RUnlock()
if d.vfs.Opt.ReadOnly {
return nil, EROFS
}
// fs.Debugf(f.Path(), "File.openWrite")
fh, err = newWriteFileHandle(d, f, f.Path(), flags)
if err != nil {
fs.Debugf(f.Path(), "File.openWrite failed: %v", err)
return nil, err
}
return fh, nil
}
// openRW open the file for read and write using a temporary file
//
// It uses the open flags passed in.
func (f *File) openRW(flags int) (fh *RWFileHandle, err error) {
f.mu.RLock()
d := f.d
f.mu.RUnlock()
// FIXME chunked
if flags&accessModeMask != os.O_RDONLY && d.vfs.Opt.ReadOnly {
return nil, EROFS
}
// fs.Debugf(f.Path(), "File.openRW")
fh, err = newRWFileHandle(d, f, flags)
if err != nil {
fs.Debugf(f.Path(), "File.openRW failed: %v", err)
return nil, err
}
return fh, nil
}
// Sync the file
//
// Note that we don't do anything except return OK
func (f *File) Sync() error {
return nil
}
// Remove the file
func (f *File) Remove() (err error) {
defer log.Trace(f.Path(), "")("err=%v", &err)
f.mu.RLock()
d := f.d
f.mu.RUnlock()
if d.vfs.Opt.ReadOnly {
return EROFS
}
// Remove the object from the cache
wasWriting := false
if d.vfs.cache != nil && d.vfs.cache.Exists(f.Path()) {
wasWriting = d.vfs.cache.Remove(f.Path())
}
// Remove the item from the directory listing
// called with File.mu released
d.delObject(f.Name())
f.muRW.Lock() // muRW must be locked before mu to avoid
f.mu.Lock() // deadlock in RWFileHandle.openPending and .close
if f.o != nil {
err = f.o.Remove(context.TODO())
}
f.mu.Unlock()
f.muRW.Unlock()
if err != nil {
if wasWriting {
// Ignore error deleting file if was writing it as it may not be uploaded yet
err = nil
fs.Debugf(f._path(), "Ignoring File.Remove file error as uploading: %v", err)
} else {
fs.Debugf(f._path(), "File.Remove file error: %v", err)
}
}
return err
}
// RemoveAll the file - same as remove for files
func (f *File) RemoveAll() error {
return f.Remove()
}
// DirEntry returns the underlying fs.DirEntry - may be nil
func (f *File) DirEntry() (entry fs.DirEntry) {
f.mu.RLock()
defer f.mu.RUnlock()
return f.o
}
// Dir returns the directory this file is in
func (f *File) Dir() *Dir {
f.mu.RLock()
defer f.mu.RUnlock()
return f.d
}
// VFS returns the instance of the VFS
func (f *File) VFS() *VFS {
f.mu.RLock()
defer f.mu.RUnlock()
return f.d.vfs
}
// Fs returns the underlying Fs for the file
func (f *File) Fs() fs.Fs {
f.mu.RLock()
defer f.mu.RUnlock()
return f.d.Fs()
}
// Open a file according to the flags provided
//
// O_RDONLY open the file read-only.
// O_WRONLY open the file write-only.
// O_RDWR open the file read-write.
//
// O_APPEND append data to the file when writing.
// O_CREATE create a new file if none exists.
// O_EXCL used with O_CREATE, file must not exist
// O_SYNC open for synchronous I/O.
// O_TRUNC if possible, truncate file when opened
//
// We ignore O_SYNC and O_EXCL
func (f *File) Open(flags int) (fd Handle, err error) {
defer log.Trace(f.Path(), "flags=%s", decodeOpenFlags(flags))("fd=%v, err=%v", &fd, &err)
var (
write bool // if set need write support
read bool // if set need read support
rdwrMode = flags & accessModeMask
)
// http://pubs.opengroup.org/onlinepubs/7908799/xsh/open.html
// The result of using O_TRUNC with O_RDONLY is undefined.
// Linux seems to truncate the file, but we prefer to return EINVAL
if rdwrMode == os.O_RDONLY && flags&os.O_TRUNC != 0 {
return nil, EINVAL
}
// Figure out the read/write intents
switch {
case rdwrMode == os.O_RDONLY:
read = true
case rdwrMode == os.O_WRONLY:
write = true
case rdwrMode == os.O_RDWR:
read = true
write = true
default:
fs.Debugf(f.Path(), "Can't figure out how to open with flags: 0x%X", flags)
return nil, EPERM
}
// If append is set then set read to force openRW
if flags&os.O_APPEND != 0 {
read = true
f.mu.Lock()
f.appendMode = true
f.mu.Unlock()
}
// If truncate is set then set write to force openRW
if flags&os.O_TRUNC != 0 {
write = true
}
// If create is set then set write to force openRW
if flags&os.O_CREATE != 0 {
write = true
}
// Open the correct sort of handle
f.mu.RLock()
d := f.d
f.mu.RUnlock()
CacheMode := d.vfs.Opt.CacheMode
if CacheMode >= vfscommon.CacheModeMinimal && (d.vfs.cache.InUse(f.Path()) || d.vfs.cache.Exists(f.Path())) {
fd, err = f.openRW(flags)
} else if read && write {
if CacheMode >= vfscommon.CacheModeMinimal {
fd, err = f.openRW(flags)
} else {
// Open write only and hope the user doesn't
// want to read. If they do they will get an
// EPERM plus an Error log.
fd, err = f.openWrite(flags)
}
} else if write {
if CacheMode >= vfscommon.CacheModeWrites {
fd, err = f.openRW(flags)
} else {
fd, err = f.openWrite(flags)
}
} else if read {
if CacheMode >= vfscommon.CacheModeFull {
fd, err = f.openRW(flags)
} else {
fd, err = f.openRead()
}
} else {
fs.Debugf(f.Path(), "Can't figure out how to open with flags: 0x%X", flags)
return nil, EPERM
}
// if creating a file, add the file to the directory
if err == nil && flags&os.O_CREATE != 0 {
// called without File.mu held
d.addObject(f)
}
return fd, err
}
// Truncate changes the size of the named file.
func (f *File) Truncate(size int64) (err error) {
// make a copy of fh.writers with the lock held then unlock so
// we can call other file methods.
f.mu.Lock()
writers := make([]Handle, len(f.writers))
copy(writers, f.writers)
f.mu.Unlock()
// If have writers then call truncate for each writer
if len(writers) != 0 {
var openWriters = len(writers)
fs.Debugf(f.Path(), "Truncating %d file handles", len(writers))
for _, h := range writers {
truncateErr := h.Truncate(size)
if truncateErr == ECLOSED {
// Ignore ECLOSED since file handle can get closed while this is running
openWriters--
} else if truncateErr != nil {
err = truncateErr
}
}
// If at least one open writer return here
if openWriters > 0 {
return err
}
}
// if o is nil it isn't valid yet
o, err := f.waitForValidObject()
if err != nil {
return err
}
// If no writers, and size is already correct then all done
if o.Size() == size {
return nil
}
fs.Debugf(f.Path(), "Truncating file")
// Otherwise if no writers then truncate the file by opening
// the file and truncating it.
flags := os.O_WRONLY
if size == 0 {
flags |= os.O_TRUNC
}
fh, err := f.Open(flags)
if err != nil {
return err
}
defer fs.CheckClose(fh, &err)
if size != 0 {
return fh.Truncate(size)
}
return nil
}