taildrop: fix theoretical race condition in fileDeleter.Init (#9876)

It is possible that upon a cold-start, we enqueue a partial file
for deletion that is resumed shortly after startup.

If the file transfer happens to last longer than deleteDelay,
we will delete the partial file, which is unfortunate.
The client spent a long time uploading a file,
only for it to be accidentally deleted.
It's a very rare race, but also a frustrating one
if it happens to manifest.

Fix the code to only delete partial files that
do not have an active puts against it.

We also fix a minor bug in ResumeReader
where we read b[:blockSize] instead of b[:cs.Size].
The former is the fixed size of 64KiB,
while the latter is usually 64KiB,
but may be less for the last block.

Updates tailscale/corp#14772

Signed-off-by: Joe Tsai <joetsai@digital-static.net>
awly/ipnlocal-watchnotifications-clientversion
Joe Tsai 1 year ago committed by GitHub
parent 25b6974219
commit 6ada33db77
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

@ -27,8 +27,8 @@ const deleteDelay = time.Hour
type fileDeleter struct {
logf logger.Logf
clock tstime.DefaultClock
event func(string) // called for certain events; for testing only
dir string
event func(string) // called for certain events; for testing only
mu sync.Mutex
queue list.List
@ -46,11 +46,11 @@ type deleteFile struct {
inserted time.Time
}
func (d *fileDeleter) Init(logf logger.Logf, clock tstime.DefaultClock, event func(string), dir string) {
d.logf = logf
d.clock = clock
d.dir = dir
d.event = event
func (d *fileDeleter) Init(m *Manager, eventHook func(string)) {
d.logf = m.opts.Logf
d.clock = m.opts.Clock
d.dir = m.opts.Dir
d.event = eventHook
// From a cold-start, load the list of partial and deleted files.
d.byName = make(map[string]*list.Element)
@ -59,19 +59,30 @@ func (d *fileDeleter) Init(logf logger.Logf, clock tstime.DefaultClock, event fu
d.group.Go(func() {
d.event("start init")
defer d.event("end init")
rangeDir(dir, func(de fs.DirEntry) bool {
rangeDir(d.dir, func(de fs.DirEntry) bool {
switch {
case d.shutdownCtx.Err() != nil:
return false // terminate early
case !de.Type().IsRegular():
return true
case strings.Contains(de.Name(), partialSuffix):
case strings.HasSuffix(de.Name(), partialSuffix):
// Only enqueue the file for deletion if there is no active put.
nameID := strings.TrimSuffix(de.Name(), partialSuffix)
if i := strings.LastIndexByte(nameID, '.'); i > 0 {
key := incomingFileKey{ClientID(nameID[i+len("."):]), nameID[:i]}
m.incomingFiles.LoadFunc(key, func(_ *incomingFile, loaded bool) {
if !loaded {
d.Insert(de.Name())
}
})
} else {
d.Insert(de.Name())
case strings.Contains(de.Name(), deletedSuffix):
}
case strings.HasSuffix(de.Name(), deletedSuffix):
// Best-effort immediate deletion of deleted files.
name := strings.TrimSuffix(de.Name(), deletedSuffix)
if os.Remove(filepath.Join(dir, name)) == nil {
if os.Remove(filepath.Join(dir, de.Name())) == nil {
if os.Remove(filepath.Join(d.dir, name)) == nil {
if os.Remove(filepath.Join(d.dir, de.Name())) == nil {
break
}
}

@ -67,8 +67,12 @@ func TestDeleter(t *testing.T) {
}
eventHook := func(event string) { eventsChan <- event }
var m Manager
var fd fileDeleter
fd.Init(t.Logf, tstime.DefaultClock{Clock: clock}, eventHook, dir)
m.opts.Logf = t.Logf
m.opts.Clock = tstime.DefaultClock{Clock: clock}
m.opts.Dir = dir
fd.Init(&m, eventHook)
defer fd.Shutdown()
insert := func(name string) {
t.Helper()

@ -145,7 +145,7 @@ func ResumeReader(r io.Reader, hashNext func() (BlockChecksum, error)) (int64, i
}
// Read the contents of the next block.
n, err := io.ReadFull(r, b[:blockSize])
n, err := io.ReadFull(r, b[:cs.Size])
b = b[:n]
if err == io.EOF || err == io.ErrUnexpectedEOF {
err = nil

@ -112,16 +112,15 @@ func (m *Manager) DeleteFile(baseName string) error {
err := os.Remove(path)
if err != nil && !os.IsNotExist(err) {
err = redactError(err)
// Put a retry loop around deletes on Windows. Windows
// file descriptor closes are effectively asynchronous,
// as a bunch of hooks run on/after close, and we can't
// necessarily delete the file for a while after close,
// as we need to wait for everybody to be done with
// it. (on Windows, unlike Unix, a file can't be deleted
// if it's open anywhere)
// So try a few times but ultimately just leave a
// "foo.jpg.deleted" marker file to note that it's
// deleted and we clean it up later.
// Put a retry loop around deletes on Windows.
//
// Windows file descriptor closes are effectively asynchronous,
// as a bunch of hooks run on/after close,
// and we can't necessarily delete the file for a while after close,
// as we need to wait for everybody to be done with it.
// On Windows, unlike Unix, a file can't be deleted if it's open anywhere.
// So try a few times but ultimately just leave a "foo.jpg.deleted"
// marker file to note that it's deleted and we clean it up later.
if runtime.GOOS == "windows" {
if bo == nil {
bo = backoff.NewBackoff("delete-retry", logf, 1*time.Second)

@ -135,7 +135,7 @@ func (opts ManagerOptions) New() *Manager {
opts.SendFileNotify = func() {}
}
m := &Manager{opts: opts}
m.deleter.Init(opts.Logf, opts.Clock, func(string) {}, opts.Dir)
m.deleter.Init(m, func(string) {})
m.emptySince.Store(-1) // invalidate this cache
return m
}
@ -250,10 +250,6 @@ func (m *Manager) IncomingFiles() []ipn.PartialFile {
return files
}
// redacted is a fake path name we use in errors, to avoid
// accidentally logging actual filenames anywhere.
const redacted = "redacted"
type redactedError struct {
msg string
inner error
@ -270,6 +266,7 @@ func (re *redactedError) Unwrap() error {
func redactString(s string) string {
hash := adler32.Checksum([]byte(s))
const redacted = "redacted"
var buf [len(redacted) + len(".12345678")]byte
b := append(buf[:0], []byte(redacted)...)
b = append(b, '.')

Loading…
Cancel
Save