windows as Source

README.md

@@ -1,9 +1,32 @@
+# WARNING
+THIS TOOL IS COMPLETELY VIBE-CODED!!!
+
+THESE LINES ARE THE ONLY ONE I HAVE WRITTEN MYSELF!!!!
+
+I SWEAR I DID NOT READ MORE THAN 20 LINES OF CODE!
+
+USE AT YOUR OWN RISK (AS ALWAYS)
+
+
+I needed a Tool to sync 2 Files from a to b with differential transfer, all managed from a workstation c.
+
+Maybe i use this to learn go...
+
+Looks nice, builds fast, runs everywhere with one Binary.
+
+
+
 # clonetool
 
 Block-level sync for a large file or block device between two machines (or
 two paths on the same machine), driven from a third, passive "manager"
 machine. Single static Go binary, no runtime dependencies beyond the
-system `ssh` client for remote endpoints.
+system `ssh` client for SSH endpoints.
+
+Runs on **Linux and Windows**. Endpoints are reached either over SSH (as
+below) or, for a host with no SSH server (typically Windows), by starting
+`clonetool listen` there and connecting to it over an encrypted, password-
+protected TCP channel — see **Listen mode (no SSH)** below.
 
 ## How it works
 
@@ -44,13 +67,25 @@ system `ssh` client for remote endpoints.
     is used at all — the source agent spawns the write-side helper as a
     plain local subprocess.
 - **Self-deploy:** if a remote endpoint has no runnable `clonetool` on
-  `PATH` (or wherever `--remote-bin` points), the manager streams *this*
+  `PATH` (or wherever `--remote-bin` points), **or the one that's there is
-  binary to `~/.clonetool/bin/clonetool` on that host over the existing
+  a different build than this binary** (see below), the manager streams
-  SSH connection and uses it — no install, no root. Disable with
+  *this* binary to `~/.clonetool/bin/clonetool` on that host over the
-  `--deploy=false`. If the copied binary won't execute there (wrong CPU
+  existing SSH connection and uses it — no install, no root. Disable with
-  architecture) the error says so; build one for the remote's arch
+  `--deploy=false` (a stale or missing binary is then an error, except a
+  merely-stale `--remote-bin` is used anyway with a warning, since there's
+  no way to correct it). If the copied binary won't execute there (wrong
+  CPU architecture) the error says so; build one for the remote's arch
   (`CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build`) and put it on `PATH`
   or pass `--remote-bin`.
+- **Version/staleness check:** every binary has a build timestamp baked in
+  (see Build below) and `clonetool version` prints it. Before using a
+  remote `clonetool`, the manager runs its `version` there and compares
+  that timestamp against its own — a mismatch (e.g. the remote was
+  self-deployed from an older build of this tool) is treated the same as
+  "not runnable" and triggers the self-deploy above, so the remote binary
+  is kept in sync with whatever you're running locally. Only the build
+  timestamp is compared, not `GOARCH` — that is expected to differ across
+  a cross-compiled deploy.
 - **sudo:** if reading or writing an endpoint that is a **block device**
   fails with a permission error, `--sudo=auto` (the default) transparently
   restarts that side's agent — and the helper it spawns on the peer —
@@ -61,67 +96,80 @@ system `ssh` client for remote endpoints.
   start; `--sudo=never` never does.
 - Sizing rules:
   - Destination is a **block device**: it can't be resized, so if the
-    source is larger the job fails; otherwise exactly `min(source, dest)`
+    source is larger the job fails; otherwise exactly the source's size is
-    bytes are synced and the remainder of the device is left untouched.
+    synced and the remainder of the device is left untouched.
   - Destination is a **regular file**: it's truncated (created if
     missing) to exactly the source's size, growing or shrinking it.
     Shrinking an existing non-empty file prompts for confirmation unless
     `--yes` is passed.
 
+## Listen mode (no SSH)
+
+For a host that has no SSH server — a Windows machine, typically — start
+clonetool there in **listen mode** and point the manager's location at it
+with a `tcp://host:port/path` spec instead of `host:path`. There is no
+self-deploy over this path: you copy the binary to that host yourself and
+start it by hand.
+
+```
+# On the source host (e.g. Windows), started manually:
+clonetool listen --bind :9000 --password s3cret
+
+# On the manager, pull that source into a local/SSH destination:
+clonetool sync --source tcp://WINHOST:9000/C:\data\disk.img \
+               --dest  backup:/srv/disk.img --password s3cret
+```
+
+- A listening host only ever **accepts** connections; it never dials out.
+  So when the **source** is a `tcp://` endpoint, clonetool always uses the
+  **pull** direction — the destination side connects to the listener and
+  pulls the changed blocks. The destination must therefore have a network
+  route to `host:port`. (Push is skipped for a listener source.)
+- **Password:** required for any `tcp://` endpoint, on both the listener
+  (`clonetool listen`) and the manager (`clonetool sync`). Provide it with
+  `--password`, `--password-file PATH`, or the `CLONETOOL_PASSWORD`
+  environment variable (env/file keep it out of the process list and shell
+  history). The manager passes it to the destination agent over the already-
+  encrypted control channel, never on a command line.
+- **Security:** the channel is TLS 1.3 (the listener uses a fresh, in-memory
+  self-signed certificate). The password is verified by a challenge-response
+  bound to the TLS session (HMAC over the connection's exported keying
+  material, key stretched with PBKDF2), so a man-in-the-middle can't
+  authenticate even though the certificate isn't checked against a CA. Both
+  ends authenticate each other; a wrong password fails the connection before
+  any file is touched. Uses only the Go standard library — still one static
+  binary.
+- The listener serves connections until you stop it with Ctrl-C. A single
+  sync opens two connections to it (one for the stat, one for the data
+  pull), so it must stay running for the whole job.
+
 ## Build
 
 ```
 CGO_ENABLED=0 go build -o clonetool .
 ```
 
-Cross-compile for another OS/arch by setting `GOOS`/`GOARCH` (no cgo, so
+Cross-compile for another architecture by setting `GOARCH` (no cgo, so
-these all work from any host):
+this works from any host):
 
 ```
 CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -o clonetool .
-CGO_ENABLED=0 GOOS=windows GOARCH=amd64 go build -o clonetool.exe    .
-CGO_ENABLED=0 GOOS=darwin  GOARCH=arm64 go build -o clonetool-darwin .
 ```
 
-`./build.sh` writes all three of the above (plus linux/amd64) into `dist/`.
+`./build.sh` writes linux/amd64 and linux/arm64 binaries into `dist/`. It
+also stamps every binary it builds with the same build timestamp (via
+`-ldflags -X main.buildTime=...`), which is how the manager tells a stale
+self-deployed remote binary apart from a current one (see "Version/staleness
+check" above) — build all your binaries for a given release with one
+`./build.sh` run so they compare equal to each other. A plain `go build`
+with no `-ldflags` leaves the timestamp at `"dev"`, which still works for
+this check (it only ever compares equal to another literal copy of the same
+`dev` binary, never to a real timestamp).
 
 Copy the resulting binary to the manager host. Source and destination
 hosts get it automatically (see self-deploy above), or place it yourself
 and point `--remote-bin` at it. The binary is architecture-specific —
-cross-compile if your hosts differ. Self-deploy streams a POSIX shell
+cross-compile if your hosts differ.
-script over SSH, so a **Windows** host can be the manager or a local
-endpoint but cannot be an automatic deploy target — put `clonetool.exe`
-on it yourself and point `--remote-bin` at it.
-
-## Windows
-
-clonetool runs on Windows and can clone **physical drives and volumes**,
-not just files:
-
-```
-# Whole disk to an image file
-clonetool sync --source \\.\PhysicalDrive2 --dest D:\backup\disk2.img
-
-# Image file back onto a disk (must not be larger than the disk)
-clonetool sync --source D:\backup\disk2.img --dest \\.\PhysicalDrive2
-
-# A single volume
-clonetool sync --source \\.\E: --dest \\.\F:
-```
-
-- Raw-disk paths are `\\.\PhysicalDrive<n>` (whole disk) or `\\.\<X>:` (a
-  volume). `\\?\` also works. Forward slashes are accepted.
-- **Run from an elevated (Administrator) console** to open a raw disk.
-  There is no `sudo` fallback on Windows; `--sudo` is ignored. A
-  permission error tells you to elevate.
-- A raw-disk **destination** should have no mounted filesystem in use
-  (take the disk offline in Disk Management, or target a volume that
-  nothing else has open) — Windows blocks writes to a disk region owned by
-  a mounted volume. Reading a live disk as the **source** is fine.
-- Raw-disk I/O must be sector-aligned. `--block-size` must be a multiple
-  of the drive's sector size (512 or 4096); the default 4M is. clonetool
-  handles the final partial block itself.
-- Device size is read with `IOCTL_DISK_GET_LENGTH_INFO`.
 
 ## Usage
 
@@ -129,43 +177,77 @@ clonetool sync --source \\.\E: --dest \\.\F:
 clonetool sync --source LOC --dest LOC [options]
 ```
 
-`LOC` is either a local path (`/dev/sdb`, `./image.bin`) or
+`LOC` is one of:
-`[user@]host:path` for a path reached over SSH.
+
+- a local path — `/dev/sdb`, `./image.bin`, `C:\data\disk.img`, `\\.\PhysicalDrive0`
+- `[user@]host:path` — reached over SSH
+- `tcp://host:port/path` — a host running `clonetool listen` (see **Listen
+  mode** above); no SSH needed there
 
 ```
 # Same machine
 clonetool sync --source /dev/sda --dest /dev/sdb
 
+# Whole disk to an image file
+clonetool sync --source /dev/sda --dest /srv/sda.img
+
 # Two remote machines, orchestrated from a third
 clonetool sync --source db1:/dev/vdb --dest backup-host:/srv/db1.img
 
+# Windows source with no SSH: it runs `clonetool listen --bind :9000 --password p`
+clonetool sync --source tcp://winbox:9000/C:\data\disk.img \
+               --dest backup-host:/srv/win.img --password p
+
 # Re-run any time; only changed blocks move
 clonetool sync --source db1:/dev/vdb --dest backup-host:/srv/db1.img
 ```
 
-Options:
+Options for `sync`:
 
 | Flag | Default | Meaning |
 |---|---|---|
 | `--block-size` | `4M` | Block size (accepts `K`/`M`/`G` suffixes). |
 | `--job` | — | Optional label shown in progress/log output. |
 | `--yes` | off | Don't prompt before shrinking an existing destination file. |
-| `--sudo` | `auto` | Block-device privilege escalation: `auto` (on a permission error), `always`, or `never`. Remote elevation needs passwordless sudo. |
+| `--sudo` | `auto` | Block-device privilege escalation: `auto` (on a permission error), `always`, or `never`. Remote elevation needs passwordless sudo. No-op on Windows (run as Administrator instead). |
-| `--deploy` | `true` | Copy this binary to remote hosts that lack a runnable `clonetool`. `--deploy=false` to disable. |
+| `--deploy` | `true` | Copy this binary to SSH hosts that lack a runnable `clonetool`. `--deploy=false` to disable. (Never applies to `tcp://` endpoints.) |
-| `--connect-timeout` | `8` | SSH connect timeout (seconds) used for the push/pull direction probe. |
+| `--connect-timeout` | `8` | SSH/TLS connect timeout (seconds) used for the push/pull direction probe. |
 | `--ssh` | `ssh` | ssh binary to use. |
 | `--ssh-opt` | — | Extra `-o OPT` passed to ssh (repeatable). |
-| `--remote-bin` | `clonetool` | Path to clonetool on remote hosts. |
+| `--remote-bin` | `clonetool` | Path to clonetool on remote SSH hosts. |
 | `--manager-host` | local hostname | Address a peer should use to reach this machine, needed only when source or dest is local to the manager *and* the other side is remote and ends up needing to dial back in (pull fallback). |
+| `--password` | — | Shared password for a `tcp://` listen endpoint. |
+| `--password-file` | — | Read the listen-endpoint password from a file (or set `CLONETOOL_PASSWORD`). |
 
-`clonetool version` prints the binary's `GOOS/GOARCH` (used internally for
+Options for `listen`:
-the self-deploy check).
+
+| Flag | Default | Meaning |
+|---|---|---|
+| `--bind` | `:9000` | Address to listen on, e.g. `:9000` or `0.0.0.0:9000`. |
+| `--password` | — | Shared password (or `--password-file`, or `CLONETOOL_PASSWORD`). Required. |
+| `--password-file` | — | Read the shared password from a file. |
+
+`clonetool version` prints the binary's `GOOS/GOARCH` and build timestamp,
+e.g. `clonetool linux/amd64 build=2024-06-01T12:00:00Z` (the timestamp is
+used internally for the self-deploy staleness check above).
+
+The status line shows read throughput for both sides and write throughput
+for the destination separately: `rd(src)` is the source reading and
+comparing its blocks, `rd(dst)` is the destination fingerprinting its
+current content (only while the trailing `scan H/M` segment is present),
+and `wr(dst)` is the destination actually writing changed blocks.
 
 ## Caveats
 
-- Block-device size detection is implemented on Linux (`BLKGETSIZE64`) and
+- Block-device sizing uses `BLKGETSIZE64` on Linux and
-  Windows (`IOCTL_DISK_GET_LENGTH_INFO`). On other systems only regular
+  `IOCTL_DISK_GET_LENGTH_INFO` on Windows (`\\.\PhysicalDrive0`, `\\.\C:`).
-  files can be synced.
+  On other platforms (e.g. macOS) only regular files are supported; a
+  device path there errors out.
+- **Windows raw disks:** syncing a `\\.\PhysicalDrive*` requires running
+  clonetool as **Administrator**, and the disk should be offline/unmounted
+  (a live, mounted volume can refuse writes or give inconsistent reads).
+  Raw-disk I/O is sector-aligned automatically; `--block-size` must be a
+  multiple of the sector size. `--sudo` does nothing on Windows.
 - If a destination path doesn't exist yet, it's created as a regular
   file — clonetool won't create device nodes, so double-check device
   paths for typos before running.

agent.go

@@ -19,20 +19,23 @@ func cmdAgent(args []string) error {
 	role := fs.String("role", "", "control|sink|source-stream (internal)")
 	path := fs.String("path", "", "path to read/write")
 	size := fs.Int64("size", 0, "total sync size in bytes")
+	base := fs.Int64("base", 0, "byte offset the window starts at")
 	blockSize := fs.Int64("block-size", defaultBlockSize, "block size in bytes")
 	if err := fs.Parse(args); err != nil {
 		return err
 	}
 
+	in := NewFrameReader(os.Stdin)
+	out := NewFrameWriter(os.Stdout)
 	switch *role {
-	case "control":
+	case roleControl:
-		return runControlAgent()
+		return runControlAgent(in, out)
-	case "sink":
+	case roleSink:
-		return runSinkRole(*path, *size, *blockSize)
+		return runSinkRole(in, out, *path, *base, *size, *blockSize)
-	case "source-stream":
+	case roleSourceStream:
-		return runSourceStreamRole(*path, *size, *blockSize)
+		return runSourceStreamRole(in, out, *path, *base, *size, *blockSize)
 	default:
-		return fmt.Errorf("agent: unknown or missing --role %q (want control|sink|source-stream)", *role)
+		return fmt.Errorf("agent: unknown or missing --role %q", *role)
 	}
 }
 
@@ -41,10 +44,7 @@ func cmdAgent(args []string) error {
 // manager over stdin/stdout with CtrlMsg frames.
 // ---------------------------------------------------------------------
 
-func runControlAgent() error {
+func runControlAgent(in *FrameReader, out *FrameWriter) error {
-	in := NewFrameReader(os.Stdin)
-	out := NewFrameWriter(os.Stdout)
-
 	for {
 		typ, payload, err := in.ReadFrame()
 		if err != nil {
@@ -125,37 +125,15 @@ func peerAgentCommand(req CtrlMsg, tailArgs []string) *exec.Cmd {
 
 func runPushDriver(req CtrlMsg, out *FrameWriter) {
 	tailArgs := []string{
-		"agent", "--role", "sink",
+		"agent", "--role", roleSink,
 		"--path", req.PeerPath,
+		"--base", strconv.FormatInt(req.PeerBase, 10),
 		"--size", strconv.FormatInt(req.Size, 10),
 		"--block-size", strconv.FormatInt(req.BlockSize, 10),
 	}
-	cmd := peerAgentCommand(req, tailArgs)
+	link, reason := openPeerLink(req, roleSink, tailArgs)
-	stdin, err := cmd.StdinPipe()
+	if reason != "" {
-	if err != nil {
+		_ = out.WriteJSON(CtrlMsg{Type: msgPushFailed, Reason: reason})
-		_ = out.WriteJSON(CtrlMsg{Type: msgPushFailed, Reason: err.Error()})
-		return
-	}
-	stdout, err := cmd.StdoutPipe()
-	if err != nil {
-		_ = out.WriteJSON(CtrlMsg{Type: msgPushFailed, Reason: err.Error()})
-		return
-	}
-	stderrBuf := newLimitedBuffer(4096)
-	cmd.Stderr = stderrBuf
-
-	if err := cmd.Start(); err != nil {
-		_ = out.WriteJSON(CtrlMsg{Type: msgPushFailed, Reason: fmt.Sprintf("start ssh: %v", err)})
-		return
-	}
-	fw := NewFrameWriter(stdin)
-	fr := NewFrameReader(stdout)
-
-	timeout := time.Duration(req.ConnectTimeoutSec+2) * time.Second
-	if err := waitReady(fr, timeout); err != nil {
-		_ = cmd.Process.Kill()
-		_ = cmd.Wait()
-		_ = out.WriteJSON(CtrlMsg{Type: msgPushFailed, Reason: fmt.Sprintf("%v (remote stderr: %s)", err, stderrBuf.String())})
 		return
 	}
 
@@ -165,26 +143,82 @@ func runPushDriver(req CtrlMsg, out *FrameWriter) {
 	// as they arrive, so the two scans and the transfer all overlap.
 	srcFile, err := os.Open(req.Path)
 	if err != nil {
-		_ = cmd.Process.Kill()
+		link.kill()
-		_ = cmd.Wait()
 		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: err.Error(), NeedPriv: isPermErr(err)})
 		return
 	}
 	defer srcFile.Close()
 
-	if fatalErr := pumpPush(req, alignmentFor(req.Path), srcFile, fw, fr, out); fatalErr != nil {
+	if fatalErr := pumpPush(req, alignmentFor(req.Path), srcFile, link.fw, link.fr, out); fatalErr != nil {
-		_ = cmd.Process.Kill()
+		link.kill()
-		_ = cmd.Wait()
 		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: fatalErr.Error()})
 		return
 	}
-	if err := cmd.Wait(); err != nil {
+	if err := link.finish(); err != nil {
-		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: fmt.Sprintf("sink process: %v (stderr: %s)", err, stderrBuf.String())})
+		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: fmt.Sprintf("sink: %v (stderr: %s)", err, link.stderr())})
 		return
 	}
 	_ = out.WriteJSON(CtrlMsg{Type: msgPushOK})
 }
 
+// peerLink is a framed, READY-confirmed connection to a peer helper (sink or
+// source-stream), reached either by spawning it over ssh/local (spawnPeerLink)
+// or by dialing a listening clonetool over TLS (dialPeerLink). It hides which
+// mechanism is in use from the push/pull drivers.
+type peerLink struct {
+	fw        *FrameWriter
+	fr        *FrameReader
+	stderr    func() string // bounded tail of the peer's stderr, "" for a net peer
+	finish    func() error  // wait for clean completion (cmd.Wait / conn close)
+	kill      func()        // force teardown on error
+	closeSend func() error  // close only the send side, to unblock a parked peer
+}
+
+// openPeerLink establishes a peerLink for role, dialing a listener when the
+// request targets a net peer and otherwise spawning the helper. A non-empty
+// reason means setup/handshake failed softly, so the caller reports
+// <push|pull>_failed and the manager can try the other direction.
+func openPeerLink(req CtrlMsg, role string, tailArgs []string) (*peerLink, string) {
+	if req.PeerNet {
+		return dialPeerLink(req, role)
+	}
+	return spawnPeerLink(req, tailArgs)
+}
+
+// spawnPeerLink runs the helper over ssh (or as a local subprocess) and waits
+// for its READY handshake.
+func spawnPeerLink(req CtrlMsg, tailArgs []string) (*peerLink, string) {
+	cmd := peerAgentCommand(req, tailArgs)
+	stdin, err := cmd.StdinPipe()
+	if err != nil {
+		return nil, err.Error()
+	}
+	stdout, err := cmd.StdoutPipe()
+	if err != nil {
+		return nil, err.Error()
+	}
+	stderrBuf := newLimitedBuffer(4096)
+	cmd.Stderr = stderrBuf
+	if err := cmd.Start(); err != nil {
+		return nil, fmt.Sprintf("start ssh: %v", err)
+	}
+	fr := NewFrameReader(stdout)
+	timeout := time.Duration(req.ConnectTimeoutSec+2) * time.Second
+	if err := waitReady(fr, timeout); err != nil {
+		_ = cmd.Process.Kill()
+		_ = cmd.Wait()
+		return nil, fmt.Sprintf("%v (remote stderr: %s)", err, stderrBuf.String())
+	}
+	return &peerLink{
+		fw:        NewFrameWriter(stdin),
+		fr:        fr,
+		stderr:    stderrBuf.String,
+		finish:    cmd.Wait,
+		kill:      func() { _ = cmd.Process.Kill(); _ = cmd.Wait() },
+		closeSend: stdin.Close,
+	}, ""
+}
+
 func waitReady(fr *FrameReader, timeout time.Duration) error {
 	type result struct {
 		typ frameType
@@ -231,7 +265,7 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 	var pendingMu sync.Mutex
 	pending := make(map[uint64]bool)
 
-	var copied, skipped int64
+	var copied, skipped, srcRead int64
 	var lastProgress time.Time
 	maybeProgress := func() {
 		if time.Since(lastProgress) < 500*time.Millisecond {
@@ -239,11 +273,24 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 		}
 		lastProgress = time.Now()
 		_ = out.WriteJSON(CtrlMsg{
-			Type: msgProgress, Copied: atomic.LoadInt64(&copied), Skipped: atomic.LoadInt64(&skipped),
+			Type: msgProgress, Phase: "xfer",
+			Copied: atomic.LoadInt64(&copied), Skipped: atomic.LoadInt64(&skipped), SrcRead: atomic.LoadInt64(&srcRead),
 			TotalBlocks: blockCount,
 		})
 	}
 
+	// lastActivityNano records the last moment either side made progress: a
+	// frame arriving from the sink (hashes, ACKs, relayed progress) or the
+	// source loop reading and comparing a block. The idle watchdog below
+	// treats a stall as fatal only when *neither* side has moved for a whole
+	// window — otherwise a check-only run, where every block matches and so no
+	// DATA frames and no ACKs are ever exchanged, trips the watchdog purely
+	// because the wire went quiet while the two scans churned through
+	// identical data.
+	var lastActivityNano int64
+	atomic.StoreInt64(&lastActivityNano, time.Now().UnixNano())
+	markActivity := func() { atomic.StoreInt64(&lastActivityNano, time.Now().UnixNano()) }
+
 	// hashCh is sized to hold every block hash so the reader below never
 	// blocks handing hashes off (which, since ACKs share the same stream,
 	// would otherwise be able to deadlock against the in-flight-send limit).
@@ -274,6 +321,7 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 				}
 				return
 			}
+			markActivity()
 			switch typ {
 			case frameBlockHash:
 				idx, h, derr := decodeBlockHashFrame(payload)
@@ -315,8 +363,10 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 	go func() {
 		sendErrCh <- runSourceLoop(sourceLoopParams{
 			File: srcFile, Size: req.Size, BlockSize: req.BlockSize, Align: align, Hashes: hashCh, Out: fw,
-			OnSkip: func(uint64) { atomic.AddInt64(&skipped, 1); maybeProgress() },
+			OnRead: func(uint64) { markActivity(); atomic.AddInt64(&srcRead, 1); maybeProgress() },
+			OnSkip: func(uint64) { markActivity(); atomic.AddInt64(&skipped, 1); maybeProgress() },
 			OnSend: func(idx uint64, _ [32]byte) {
+				markActivity()
 				sem <- struct{}{}
 				pendingMu.Lock()
 				pending[idx] = true
@@ -332,7 +382,11 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 	sendCh := sendErrCh
 	ackCh := ackEvents
 	var fatalErr error
-	idle := time.NewTimer(120 * time.Second)
+	// idleTimeout is how long *both* sides can be silent before the transfer
+	// is declared stalled. The timer may fire sooner; a fire is only fatal
+	// when lastActivityNano confirms nothing has moved for the whole window.
+	const idleTimeout = 120 * time.Second
+	idle := time.NewTimer(idleTimeout)
 	defer idle.Stop()
 	for sendCh != nil || ackCh != nil {
 		if !idle.Stop() {
@@ -341,7 +395,7 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 			default:
 			}
 		}
-		idle.Reset(120 * time.Second)
+		idle.Reset(idleTimeout)
 		select {
 		case sendErr := <-sendCh:
 			sendCh = nil
@@ -367,7 +421,11 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 				maybeProgress()
 			}
 		case <-idle.C:
-			fatalErr = fmt.Errorf("timed out waiting for the sink")
+			idleFor := time.Since(time.Unix(0, atomic.LoadInt64(&lastActivityNano)))
+			if idleFor < idleTimeout {
+				continue // progress somewhere within the window; re-arm and wait
+			}
+			fatalErr = fmt.Errorf("timed out waiting for the sink (no activity for %s)", idleFor.Round(time.Second))
 		}
 		if fatalErr != nil {
 			break
@@ -384,7 +442,11 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 		return fmt.Errorf("sink closed the connection with %d block write confirmation(s) still outstanding", n)
 	}
 
-	_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Copied: atomic.LoadInt64(&copied), Skipped: atomic.LoadInt64(&skipped), TotalBlocks: blockCount})
+	_ = out.WriteJSON(CtrlMsg{
+		Type: msgProgress, Phase: "xfer",
+		Copied: atomic.LoadInt64(&copied), Skipped: atomic.LoadInt64(&skipped), SrcRead: atomic.LoadInt64(&srcRead),
+		TotalBlocks: blockCount,
+	})
 	return nil
 }
 
@@ -398,39 +460,19 @@ func pumpPush(req CtrlMsg, align int64, srcFile *os.File, fw *FrameWriter, fr *F
 
 func runPullDriver(req CtrlMsg, out *FrameWriter) {
 	tailArgs := []string{
-		"agent", "--role", "source-stream",
+		"agent", "--role", roleSourceStream,
 		"--path", req.PeerPath,
+		"--base", strconv.FormatInt(req.PeerBase, 10),
 		"--size", strconv.FormatInt(req.Size, 10),
 		"--block-size", strconv.FormatInt(req.BlockSize, 10),
 	}
-	cmd := peerAgentCommand(req, tailArgs)
+	link, reason := openPeerLink(req, roleSourceStream, tailArgs)
-	stdin, err := cmd.StdinPipe()
+	if reason != "" {
-	if err != nil {
+		_ = out.WriteJSON(CtrlMsg{Type: msgPullFailed, Reason: reason})
-		_ = out.WriteJSON(CtrlMsg{Type: msgPullFailed, Reason: err.Error()})
-		return
-	}
-	stdout, err := cmd.StdoutPipe()
-	if err != nil {
-		_ = out.WriteJSON(CtrlMsg{Type: msgPullFailed, Reason: err.Error()})
-		return
-	}
-	stderrBuf := newLimitedBuffer(4096)
-	cmd.Stderr = stderrBuf
-
-	if err := cmd.Start(); err != nil {
-		_ = out.WriteJSON(CtrlMsg{Type: msgPullFailed, Reason: fmt.Sprintf("start ssh: %v", err)})
-		return
-	}
-	fw := NewFrameWriter(stdin)
-	fr := NewFrameReader(stdout)
-
-	timeout := time.Duration(req.ConnectTimeoutSec+2) * time.Second
-	if err := waitReady(fr, timeout); err != nil {
-		_ = cmd.Process.Kill()
-		_ = cmd.Wait()
-		_ = out.WriteJSON(CtrlMsg{Type: msgPullFailed, Reason: fmt.Sprintf("%v (remote stderr: %s)", err, stderrBuf.String())})
 		return
 	}
+	fw := link.fw
+	fr := link.fr
 
 	// Handshake succeeded. Open the destination and fingerprint its current
 	// content block by block, streaming each hash to the source stream the
@@ -438,8 +480,7 @@ func runPullDriver(req CtrlMsg, out *FrameWriter) {
 	// whatever it streams back into the same file concurrently.
 	dstFile, err := os.OpenFile(req.Path, os.O_RDWR, 0)
 	if err != nil {
-		_ = cmd.Process.Kill()
+		link.kill()
-		_ = cmd.Wait()
 		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: err.Error(), NeedPriv: isPermErr(err)})
 		return
 	}
@@ -448,7 +489,7 @@ func runPullDriver(req CtrlMsg, out *FrameWriter) {
 
 	hashErrCh := make(chan error, 1)
 	go func() {
-		err := streamHashBlocks(dstFile, req.Size, req.BlockSize, align, func(bh blockHash) error {
+		err := streamHashBlocks(dstFile, req.Base, req.Size, req.BlockSize, align, func(bh blockHash) error {
 			return fw.WriteFrame(frameBlockHash, encodeBlockHashFrame(bh.index, bh.hash))
 		}, scanProgressEmitter(out))
 		if err == nil {
@@ -457,35 +498,43 @@ func runPullDriver(req CtrlMsg, out *FrameWriter) {
 		if err != nil {
 			// Unblock the source stream (waiting for more hashes) so the
 			// dest loop below can unwind instead of hanging.
-			_ = stdin.Close()
+			_ = link.closeSend()
 		}
 		hashErrCh <- err
 	}()
 
+	var written int64
+	var lastWritten time.Time
 	loopErr := runDestLoop(destLoopParams{
-		File: dstFile, BlockSize: req.BlockSize, Align: align, In: fr,
+		File: dstFile, Base: req.Base, BlockSize: req.BlockSize, Align: align, In: fr,
 		OnCtrlMsg: func(m CtrlMsg) {
 			if m.Type == msgProgress {
 				_ = out.WriteJSON(m)
 			}
 		},
+		OnWrite: func(uint64) {
+			written++
+			if time.Since(lastWritten) >= 500*time.Millisecond {
+				lastWritten = time.Now()
+				_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "write", Written: written})
+			}
+		},
 	})
+	_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "write", Written: written})
 	hashErr := <-hashErrCh
 
 	if hashErr != nil {
-		_ = cmd.Process.Kill()
+		link.kill()
-		_ = cmd.Wait()
 		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: fmt.Sprintf("scan destination: %v", hashErr), NeedPriv: isPermErr(hashErr)})
 		return
 	}
 	if loopErr != nil {
-		_ = cmd.Process.Kill()
+		link.kill()
-		_ = cmd.Wait()
 		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: loopErr.Error()})
 		return
 	}
-	if err := cmd.Wait(); err != nil {
+	if err := link.finish(); err != nil {
-		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: fmt.Sprintf("source-stream process: %v (stderr: %s)", err, stderrBuf.String())})
+		_ = out.WriteJSON(CtrlMsg{Type: msgError, Message: fmt.Sprintf("source-stream: %v (stderr: %s)", err, link.stderr())})
 		return
 	}
 	_ = out.WriteJSON(CtrlMsg{Type: msgPullOK})
@@ -496,10 +545,7 @@ func runPullDriver(req CtrlMsg, out *FrameWriter) {
 // destination host. Dumb write endpoint: verify+pwrite+ack per block.
 // ---------------------------------------------------------------------
 
-func runSinkRole(path string, size, blockSize int64) error {
+func runSinkRole(in *FrameReader, out *FrameWriter, path string, base, size, blockSize int64) error {
-	out := NewFrameWriter(os.Stdout)
-	in := NewFrameReader(os.Stdin)
-
 	f, err := os.OpenFile(path, os.O_RDWR, 0)
 	if err != nil {
 		fmt.Fprintf(os.Stderr, "sink: open %s: %v\n", path, err)
@@ -519,11 +565,33 @@ func runSinkRole(path string, size, blockSize int64) error {
 	// block is always hashed before a write for it can arrive (the source
 	// only sends after seeing that block's hash), so the two accesses to f
 	// never race on the same region.
+	// While the destination scan runs, emit a heartbeat on a fixed cadence
+	// even if streamHashBlocks is parked inside one slow ReadAt (a big /
+	// non-sparse / remote target). The push driver's idle watchdog counts any
+	// frame from the sink as activity, so this stops a legitimately slow scan
+	// from being declared a stalled sink during a check-only run, where there
+	// is no DATA/ACK traffic to feed the watchdog. A sink that has actually
+	// died stops heartbeating and its pipe closes, so the watchdog still fires.
+	scanDone := make(chan struct{})
+	go func() {
+		t := time.NewTicker(10 * time.Second)
+		defer t.Stop()
+		for {
+			select {
+			case <-scanDone:
+				return
+			case <-t.C:
+				_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "scan"})
+			}
+		}
+	}()
+
 	hashErrCh := make(chan error, 1)
 	go func() {
-		err := streamHashBlocks(f, size, blockSize, align, func(bh blockHash) error {
+		err := streamHashBlocks(f, base, size, blockSize, align, func(bh blockHash) error {
 			return out.WriteFrame(frameBlockHash, encodeBlockHashFrame(bh.index, bh.hash))
 		}, scanProgressEmitter(out))
+		close(scanDone)
 		if err == nil {
 			err = out.WriteFrame(frameHashDone, nil)
 		} else {
@@ -533,7 +601,19 @@ func runSinkRole(path string, size, blockSize int64) error {
 		hashErrCh <- err
 	}()
 
-	loopErr := runDestLoop(destLoopParams{File: f, BlockSize: blockSize, Align: align, In: in, AckOut: out})
+	var written int64
+	var lastWritten time.Time
+	loopErr := runDestLoop(destLoopParams{
+		File: f, BlockSize: blockSize, Align: align, In: in, AckOut: out,
+		OnWrite: func(uint64) {
+			written++
+			if time.Since(lastWritten) >= 500*time.Millisecond {
+				lastWritten = time.Now()
+				_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "write", Written: written})
+			}
+		},
+	})
+	_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "write", Written: written})
 	if hashErr := <-hashErrCh; hashErr != nil {
 		return hashErr
 	}
@@ -547,10 +627,7 @@ func runSinkRole(path string, size, blockSize int64) error {
 // writing straight to its own stdout.
 // ---------------------------------------------------------------------
 
-func runSourceStreamRole(path string, size, blockSize int64) error {
+func runSourceStreamRole(in *FrameReader, out *FrameWriter, path string, base, size, blockSize int64) error {
-	out := NewFrameWriter(os.Stdout)
-	in := NewFrameReader(os.Stdin)
-
 	f, err := os.Open(path)
 	if err != nil {
 		fmt.Fprintf(os.Stderr, "source-stream: open %s: %v\n", path, err)
@@ -599,21 +676,23 @@ func runSourceStreamRole(path string, size, blockSize int64) error {
 		}
 	}()
 
-	var copied, skipped int64
+	var copied, skipped, srcRead int64
 	var lastProgress time.Time
 	maybeProgress := func() {
 		if time.Since(lastProgress) < 500*time.Millisecond {
 			return
 		}
 		lastProgress = time.Now()
-		_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Copied: copied, Skipped: skipped, TotalBlocks: blockCount})
+		_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "xfer", Copied: copied, Skipped: skipped, SrcRead: srcRead, TotalBlocks: blockCount})
 	}
 
 	loopErr := runSourceLoop(sourceLoopParams{
 		File: f, Size: size, BlockSize: blockSize, Align: align, Hashes: hashCh, Out: out,
+		OnRead: func(uint64) { srcRead++; maybeProgress() },
 		OnSkip: func(uint64) { skipped++; maybeProgress() },
 		OnSend: func(uint64, [32]byte) { copied++; maybeProgress() },
 	})
+	_ = out.WriteJSON(CtrlMsg{Type: msgProgress, Phase: "xfer", Copied: copied, Skipped: skipped, SrcRead: srcRead, TotalBlocks: blockCount})
 	if readErr := <-readErrCh; readErr != nil {
 		return readErr
 	}

build.sh

@@ -4,18 +4,19 @@ set -e
 cd "$(dirname "$0")"
 mkdir -p dist
 
+build_time=$(date -u +%Y-%m-%dT%H:%M:%SZ)
+
 build() {
 	os=$1 arch=$2 out=$3
 	echo "==> $os/$arch -> dist/$out"
-	CGO_ENABLED=0 GOOS="$os" GOARCH="$arch" go build -trimpath -ldflags "-s -w" -o "dist/$out" .
+	CGO_ENABLED=0 GOOS="$os" GOARCH="$arch" go build -trimpath \
+		-ldflags "-s -w -X main.buildTime=$build_time" -o "dist/$out" .
 }
 
 build linux   amd64 clonetool-linux-amd64
 build linux   arm64 clonetool-linux-arm64
 build windows amd64 clonetool-windows-amd64.exe
 build windows arm64 clonetool-windows-arm64.exe
-build darwin  amd64 clonetool-darwin-amd64
-build darwin  arm64 clonetool-darwin-arm64
 
 echo "done:"
 ls -la dist/

control.go

@@ -6,8 +6,11 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"net"
 	"os"
 	"os/exec"
+	"strconv"
+	"time"
 )
 
 // errNeedPriv is wrapped into the error from a control-agent call that
@@ -187,6 +190,9 @@ func (c *Controller) connectAndPump(req CtrlMsg, cb transferCallbacks, okType, f
 func (c *Controller) Close() error {
 	_ = c.fw.WriteJSON(CtrlMsg{Type: msgClose})
 	_ = c.in.Close()
+	if c.cmd == nil {
+		return nil // net transport: closing the connection is enough
+	}
 	err := c.cmd.Wait()
 	if err != nil {
 		var exitErr *exec.ExitError
@@ -197,3 +203,22 @@ func (c *Controller) Close() error {
 	}
 	return nil
 }
+
+// startNetworkController connects to a listening clonetool (spec must be a net
+// endpoint) over TLS, authenticates with password, and requests the control
+// role. The returned Controller behaves like an ssh/local one for the manager;
+// it has no child process (cmd == nil).
+func startNetworkController(spec Spec, tag, password string, connectTimeoutSec int) (*Controller, error) {
+	_ = tag // errors are prefixed by the caller (source:/dest:)
+	addr := net.JoinHostPort(spec.Host, strconv.Itoa(spec.Port))
+	timeout := time.Duration(connectTimeoutSec+2) * time.Second
+	conn, fr, fw, err := dialAuth(addr, password, timeout)
+	if err != nil {
+		return nil, fmt.Errorf("connect to listener %s: %w", addr, err)
+	}
+	if err := fw.WriteJSON(CtrlMsg{Type: msgRole, Role: roleControl}); err != nil {
+		_ = conn.Close()
+		return nil, fmt.Errorf("request control role on %s: %w", addr, err)
+	}
+	return &Controller{tag: tag, in: conn, fw: fw, fr: fr}, nil
+}

ctrlmsg.go

@@ -19,8 +19,12 @@ type CtrlMsg struct {
 	// connect_push (-> source agent) / connect_pull (-> dest agent)
 	PeerHost string `json:"peerHost,omitempty"`
 	PeerUser string `json:"peerUser,omitempty"`
-	PeerPort int    `json:"peerPort,omitempty"`
 	PeerPath string `json:"peerPath,omitempty"`
+	// Base/PeerBase shift every block offset on the local/peer side. Zero on
+	// both sides (the whole-file case) is what `sync` uses; the sink and
+	// source-stream roles still accept a non-zero base for a windowed copy.
+	Base     int64 `json:"base,omitempty"`
+	PeerBase int64 `json:"peerBase,omitempty"`
 	// PeerLocal is set when both source and dest are local to the manager,
 	// so the driver (itself already a local child of the manager) can spawn
 	// the sink/source-stream helper as a plain local subprocess instead of
@@ -34,23 +38,41 @@ type CtrlMsg struct {
 	// Sudo tells a push/pull driver to run the peer helper it spawns
 	// (sink / source-stream) under "sudo -n" (remote) or "sudo" (local).
 	Sudo bool `json:"sudo,omitempty"`
+	// Peer listen endpoint: when PeerNet is set, the push/pull driver reaches
+	// the peer by dialing a listening clonetool over TLS (authenticated with
+	// PeerPassword) instead of spawning it over ssh. PeerHost carries the
+	// listener host in that case; PeerPort its port. PeerPassword travels only
+	// over the manager<->control channel (ssh-encrypted or TLS), never argv.
+	PeerNet      bool   `json:"peerNet,omitempty"`
+	PeerPort     int    `json:"peerPort,omitempty"`
+	PeerPassword string `json:"peerPassword,omitempty"`
+
+	// Role is the endpoint requested on a listen connection's role-request
+	// frame (control | sink | source-stream), see nettransport.go / listen.go.
+	Role string `json:"role,omitempty"`
 
 	// failure/error detail
 	Reason  string `json:"reason,omitempty"`
 	Message string `json:"message,omitempty"`
 
-	// progress. Scan and transfer now overlap, so a run emits both "scan"
+	// progress. Scan, source read and destination write all overlap and are
-	// messages (Hashed/TotalBlocks) and transfer messages (Copied/Skipped)
+	// tracked independently, so a run emits several kinds of "progress"
-	// interleaved; the printer keeps the latest of each.
+	// message distinguished by Phase, interleaved; the printer keeps the
-	Phase       string `json:"phase,omitempty"` // "scan" while hashing the destination; empty during transfer
+	// latest of each:
+	//   "scan"  - Hashed/TotalBlocks: destination read throughput while it
+	//             fingerprints its current content.
+	//   "xfer"  - Copied/Skipped/SrcRead/TotalBlocks: source read
+	//             throughput as it reads and compares each block (Skipped:
+	//             matched the destination; Copied: differed and was sent).
+	//   "write" - Written: destination write throughput as changed blocks
+	//             are actually applied.
+	Phase       string `json:"phase,omitempty"`
 	Hashed      int64  `json:"hashed,omitempty"`
 	Copied      int64  `json:"copied,omitempty"`
 	Skipped     int64  `json:"skipped,omitempty"`
+	SrcRead     int64  `json:"srcRead,omitempty"`
+	Written     int64  `json:"written,omitempty"`
 	TotalBlocks int64  `json:"totalBlocks,omitempty"`
-	BytesCopied int64  `json:"bytesCopied,omitempty"`
-
-	// log
-	Level string `json:"level,omitempty"`
 }
 
 const (
@@ -65,8 +87,19 @@ const (
 	msgPullOK      = "pull_ok"
 	msgPullFailed  = "pull_failed"
 	msgProgress    = "progress"
-	msgLog         = "log"
 	msgError       = "error"
 	msgClose       = "close"
 	msgBye         = "bye"
+	// msgRole is the first frame on a listen connection after authentication:
+	// it names the role to run (control | sink | source-stream) and, for the
+	// one-shot data roles, the path/window to operate on.
+	msgRole    = "role"
+	msgRoleOK  = "role_ok"
+	msgRoleErr = "role_err"
+)
+
+const (
+	roleControl      = "control"
+	roleSink         = "sink"
+	roleSourceStream = "source-stream"
 )

deploy.go

@@ -15,26 +15,43 @@ import (
 // no install, no root.
 const deployedRemoteBin = ".clonetool/bin/clonetool"
 
-// resolveRemoteBin returns the path to a runnable clonetool on spec's host.
+// resolveRemoteBin returns the path to a clonetool on spec's host that is
-// It checks the configured --remote-bin first, then a copy deployed by an
+// both runnable and the same build as this binary (compared via the
-// earlier run, and finally streams this binary over and verifies it runs.
+// "build=" tag in `version` output, see versionString/remoteBuildTag — an
+// architecture mismatch alone, expected when cross-compiled, doesn't count
+// as a mismatch). It checks the configured --remote-bin first, then a copy
+// deployed by an earlier run, and finally streams this binary over and
+// verifies it runs. With --deploy=false it can't correct a stale or missing
+// binary, so it uses --remote-bin anyway if merely out of date (only a
+// non-runnable one is still an error).
 func resolveRemoteBin(cfg *SyncConfig, spec Spec, tag string) (string, error) {
-	if _, code, err := runRemote(cfg, spec, cfg.RemoteBin, "version"); err != nil {
+	localTag := remoteBuildTag(versionString())
+
+	if out, code, err := runRemote(cfg, spec, cfg.RemoteBin, "version"); err != nil {
 		return "", fmt.Errorf("%s: cannot reach %s over ssh: %w", tag, spec.Host, err)
 	} else if code == 0 {
+		if remoteBuildTag(out) == localTag {
 			return cfg.RemoteBin, nil
 		}
+		if !cfg.Deploy {
+			fmt.Fprintf(os.Stderr, "%s: %s on %s reports %q, a different build than this one (%s); continuing anyway (--deploy=false)\n",
+				tag, cfg.RemoteBin, spec.Host, firstLine(out), versionString())
+			return cfg.RemoteBin, nil
+		}
+		fmt.Fprintf(os.Stderr, "%s: %s on %s reports %q, a different build than this one (%s); looking for an up-to-date copy ...\n",
+			tag, cfg.RemoteBin, spec.Host, firstLine(out), versionString())
+	}
 
 	if !cfg.Deploy {
 		return "", fmt.Errorf("%s: %q is not runnable on %s and --deploy=false; "+
 			"install clonetool there or pass --remote-bin", tag, cfg.RemoteBin, spec.Host)
 	}
 
-	if _, code, err := runRemote(cfg, spec, deployedRemoteBin, "version"); err == nil && code == 0 {
+	if out, code, err := runRemote(cfg, spec, deployedRemoteBin, "version"); err == nil && code == 0 && remoteBuildTag(out) == localTag {
 		return deployedRemoteBin, nil
 	}
 
-	fmt.Fprintf(os.Stderr, "%s: no runnable clonetool on %s; copying this binary there ...\n", tag, spec.Host)
+	fmt.Fprintf(os.Stderr, "%s: no up-to-date clonetool on %s; copying this binary there ...\n", tag, spec.Host)
 	if err := deployBinary(cfg, spec); err != nil {
 		return "", fmt.Errorf("%s: copy clonetool to %s: %w", tag, spec.Host, err)
 	}

device.go

@@ -13,20 +13,22 @@ type PathInfo struct {
 	Size     int64
 }
 
+// alignmentFor returns the offset/length alignment a path's handle requires
+// for positioned reads and writes. On Linux this is always 1 (block devices
+// accept buffered pread/pwrite at any alignment); on Windows a raw disk handle
+// requires sector-aligned I/O, so deviceAlignment reports the sector size.
+func alignmentFor(path string) int64 { return deviceAlignment(path) }
+
 func statPath(path string) (PathInfo, error) {
-	// A raw disk handle (Windows \\.\PhysicalDrive0, \\.\C:) is not something
+	// Windows raw-disk paths (\\.\PhysicalDrive0) aren't reported as devices by
-	// os.Stat can describe, so ask the platform for its size directly.
+	// os.Stat, so recognize them by name first and size them via ioctl.
-	if isDevicePath(path) {
+	if isRawDevice(path) {
-		size, err := blockDeviceSize(path)
+		sz, err := blockDeviceSize(path)
 		if err != nil {
-			if os.IsNotExist(err) {
-				return PathInfo{Exists: false}, nil
-			}
 			return PathInfo{}, fmt.Errorf("stat device %s: %w", path, err)
 		}
-		return PathInfo{Exists: true, IsDevice: true, Size: size}, nil
+		return PathInfo{Exists: true, IsDevice: true, Size: sz}, nil
 	}
-
 	fi, err := os.Stat(path)
 	if err != nil {
 		if os.IsNotExist(err) {
@@ -63,9 +65,6 @@ func prepareDest(path string, targetSize int64) error {
 		}
 		return nil
 	}
-	if isDevicePath(path) && !info.Exists {
-		return fmt.Errorf("destination device %s not found", path)
-	}
 	f, err := os.OpenFile(path, os.O_CREATE|os.O_WRONLY, 0644)
 	if err != nil {
 		return fmt.Errorf("open/create %s: %w", path, err)

device_linux.go

@@ -26,3 +26,16 @@ func blockDeviceSize(path string) (int64, error) {
 	}
 	return int64(size), nil
 }
+
+// isRawDevice reports paths that must be sized via ioctl rather than os.Stat.
+// On Linux block devices are recognized through os.ModeDevice instead, so this
+// is always false.
+func isRawDevice(string) bool { return false }
+
+// deviceAlignment is 1 on Linux: buffered pread/pwrite on a block device accept
+// any alignment.
+func deviceAlignment(string) int64 { return 1 }
+
+// canElevate reports whether a permission failure opening a device is worth
+// retrying under sudo (see --sudo).
+func canElevate() bool { return true }

device_other.go

@@ -2,8 +2,21 @@
 
 package main
 
-import "fmt"
+import (
+	"fmt"
+	"runtime"
+)
 
-func blockDeviceSize(path string) (int64, error) {
+// Stubs so clonetool still builds (and runs for regular-file syncs) on
-	return 0, fmt.Errorf("block device size detection is only implemented on linux and windows (got path %s)", path)
+// platforms without raw block-device support wired up (e.g. macOS for local
+// development). Block-device paths error out clearly.
+
+func isRawDevice(string) bool { return false }
+
+func blockDeviceSize(string) (int64, error) {
+	return 0, fmt.Errorf("block devices are not supported on %s", runtime.GOOS)
 }
+
+func deviceAlignment(string) int64 { return 1 }
+
+func canElevate() bool { return true }

device_unix.go

@@ -1,18 +0,0 @@
-//go:build !windows
-
-package main
-
-// isDevicePath reports whether a path must be treated as a raw device from
-// its string form alone. On unix, os.Stat's mode bits already identify
-// device nodes accurately (symlinks, non-standard locations and all), so
-// this is always false and statPath relies on those instead.
-func isDevicePath(string) bool { return false }
-
-// alignmentFor returns the offset/length alignment a path's handle requires
-// for positioned reads and writes. Unix block devices accept ordinary
-// buffered pread/pwrite at any alignment, so there is nothing to round to.
-func alignmentFor(string) int64 { return 1 }
-
-// canElevate reports whether a permission failure opening a device is worth
-// retrying under `sudo` (see --sudo). Always true on unix.
-func canElevate() bool { return true }

device_windows.go

@@ -3,97 +3,96 @@
 package main
 
 import (
-	"fmt"
+	"encoding/binary"
-	"os"
 	"strings"
 	"syscall"
-	"unsafe"
 )
 
-// isDevicePath recognises the Win32 device namespaces clonetool can sync
+// IOCTL codes (winioctl.h):
-// against by string form: \\.\PhysicalDrive0 (a whole disk) and \\.\C: (a
+//
-// volume). Both the \\.\ and \\?\ prefixes, and their forward-slash
+//	IOCTL_DISK_GET_LENGTH_INFO    -> GET_LENGTH_INFORMATION { LARGE_INTEGER Length }
-// spellings, are accepted. os.Stat can't describe these paths, so statPath
+//	IOCTL_DISK_GET_DRIVE_GEOMETRY -> DISK_GEOMETRY (BytesPerSector at offset 20)
-// keys off this instead of mode bits on Windows.
-func isDevicePath(path string) bool {
-	p := strings.ReplaceAll(path, "/", `\`)
-	return strings.HasPrefix(p, `\\.\`) || strings.HasPrefix(p, `\\?\`)
-}
-
-// canElevate is false on Windows: there is no `sudo` equivalent to re-exec
-// under, so a permission failure opening a raw disk is reported with a hint
-// to run from an elevated console instead of being retried.
-func canElevate() bool { return false }
-
 const (
-	ioctlDiskGetLengthInfo    = 0x0007405C // IOCTL_DISK_GET_LENGTH_INFO
+	ioctlDiskGetLengthInfo    = 0x0007405C
-	ioctlDiskGetDriveGeometry = 0x00070000 // IOCTL_DISK_GET_DRIVE_GEOMETRY
+	ioctlDiskGetDriveGeometry = 0x00070000
 )
 
-// diskGeometry mirrors DISK_GEOMETRY (24 bytes; Cylinders is a LARGE_INTEGER).
+// isRawDevice recognizes the raw-disk paths that os.Stat won't report as
-type diskGeometry struct {
+// devices: \\.\PhysicalDrive<N> and the volume form \\.\X:.
-	Cylinders         int64
+func isRawDevice(path string) bool {
-	MediaType         uint32
+	l := strings.ToLower(path)
-	TracksPerCylinder uint32
+	if strings.HasPrefix(l, `\\.\physicaldrive`) {
-	SectorsPerTrack   uint32
+		return true
-	BytesPerSector    uint32
+	}
+	// \\.\C: — a single drive letter behind the device namespace.
+	if strings.HasPrefix(path, `\\.\`) && len(path) == 6 && path[5] == ':' {
+		return true
+	}
+	return false
 }
 
-// openDeviceHandle opens path for a metadata ioctl only: zero access rights
+// openDevice opens a raw device handle with the sharing a live disk needs.
-// (which need no privilege and don't require the volume to be unlocked) and
+func openDevice(path string, write bool) (syscall.Handle, error) {
-// shared read/write so it doesn't disturb a mounted filesystem.
-func openDeviceHandle(path string) (syscall.Handle, error) {
 	p, err := syscall.UTF16PtrFromString(path)
 	if err != nil {
 		return syscall.InvalidHandle, err
 	}
-	return syscall.CreateFile(p, 0,
+	access := uint32(syscall.GENERIC_READ)
+	if write {
+		access |= syscall.GENERIC_WRITE
+	}
+	return syscall.CreateFile(p, access,
 		syscall.FILE_SHARE_READ|syscall.FILE_SHARE_WRITE, nil,
 		syscall.OPEN_EXISTING, 0, 0)
 }
 
-// blockDeviceSize returns the byte length of a physical drive or volume via
-// DeviceIoControl(IOCTL_DISK_GET_LENGTH_INFO).
 func blockDeviceSize(path string) (int64, error) {
-	h, err := openDeviceHandle(path)
+	h, err := openDevice(path, false)
 	if err != nil {
-		return 0, &os.PathError{Op: "open", Path: path, Err: err}
+		return 0, err
 	}
 	defer syscall.CloseHandle(h)
 
-	var length int64 // GET_LENGTH_INFORMATION is a single LARGE_INTEGER
+	var out [8]byte
 	var ret uint32
-	err = syscall.DeviceIoControl(h, ioctlDiskGetLengthInfo,
+	if err := syscall.DeviceIoControl(h, ioctlDiskGetLengthInfo, nil, 0,
-		nil, 0,
+		&out[0], uint32(len(out)), &ret, nil); err != nil {
-		(*byte)(unsafe.Pointer(&length)), uint32(unsafe.Sizeof(length)),
+		return 0, err
-		&ret, nil)
-	if err != nil {
-		return 0, fmt.Errorf("IOCTL_DISK_GET_LENGTH_INFO %s: %w", path, err)
 	}
-	return length, nil
+	return int64(binary.LittleEndian.Uint64(out[:])), nil
 }
 
-// alignmentFor returns the sector size a raw disk handle's positioned reads
+// deviceSectorSize queries the disk's physical sector size (bytes-per-sector),
-// and writes must be aligned to; 1 for an ordinary file path. It falls back
+// falling back to 512 if the geometry can't be read.
-// to 512 if the geometry query fails.
+func deviceSectorSize(path string) int64 {
-func alignmentFor(path string) int64 {
+	h, err := openDevice(path, false)
-	if !isDevicePath(path) {
+	if err != nil {
+		return 512
+	}
+	defer syscall.CloseHandle(h)
+
+	var geom [24]byte
+	var ret uint32
+	if err := syscall.DeviceIoControl(h, ioctlDiskGetDriveGeometry, nil, 0,
+		&geom[0], uint32(len(geom)), &ret, nil); err != nil {
+		return 512
+	}
+	bps := binary.LittleEndian.Uint32(geom[20:24])
+	if bps == 0 {
+		return 512
+	}
+	return int64(bps)
+}
+
+// deviceAlignment reports the sector alignment a raw disk handle requires;
+// regular files need none.
+func deviceAlignment(path string) int64 {
+	if isRawDevice(path) {
+		return deviceSectorSize(path)
+	}
 	return 1
 }
-	h, err := openDeviceHandle(path)
-	if err != nil {
-		return 512
-	}
-	defer syscall.CloseHandle(h)
 
-	var g diskGeometry
+// canElevate is false on Windows: there is no sudo. Raw-disk access requires
-	var ret uint32
+// running clonetool as Administrator, which the operator arranges outside the tool.
-	err = syscall.DeviceIoControl(h, ioctlDiskGetDriveGeometry,
+func canElevate() bool { return false }
-		nil, 0,
-		(*byte)(unsafe.Pointer(&g)), uint32(unsafe.Sizeof(g)),
-		&ret, nil)
-	if err != nil || g.BytesPerSector == 0 {
-		return 512
-	}
-	return int64(g.BytesPerSector)
-}

listen.go

@@ -0,0 +1,122 @@
+package main
+
+import (
+	"crypto/tls"
+	"flag"
+	"fmt"
+	"net"
+	"os"
+	"strings"
+	"time"
+)
+
+// cmdListen runs clonetool in listen mode: it waits for inbound TLS
+// connections (see nettransport.go), authenticates each with the shared
+// password, and serves whatever role the peer requests (control for the
+// manager's stat calls, source-stream/sink for the data transfer). This is the
+// no-ssh path for a host — typically a Windows source — that this machine and
+// the destination connect *to* instead of being reached over ssh.
+func cmdListen(args []string) error {
+	fs := flag.NewFlagSet("listen", flag.ContinueOnError)
+	bind := fs.String("bind", ":9000", "address to listen on, e.g. :9000 or 0.0.0.0:9000")
+	password := fs.String("password", "", "shared password (or set CLONETOOL_PASSWORD, or --password-file)")
+	passwordFile := fs.String("password-file", "", "read the shared password from this file")
+	if err := fs.Parse(args); err != nil {
+		return err
+	}
+	pw, err := resolvePassword(*password, *passwordFile)
+	if err != nil {
+		return err
+	}
+	if pw == "" {
+		return fmt.Errorf("listen: a password is required (--password, --password-file, or CLONETOOL_PASSWORD)")
+	}
+
+	tlsCfg, err := newServerTLSConfig()
+	if err != nil {
+		return fmt.Errorf("listen: tls setup: %w", err)
+	}
+	ln, err := net.Listen("tcp", *bind)
+	if err != nil {
+		return fmt.Errorf("listen on %s: %w", *bind, err)
+	}
+	defer ln.Close()
+	fmt.Fprintf(os.Stderr, "clonetool listening on %s (%s); press Ctrl-C to stop\n", ln.Addr(), versionString())
+
+	for {
+		raw, err := ln.Accept()
+		if err != nil {
+			return fmt.Errorf("accept: %w", err)
+		}
+		go serveListenConn(raw, tlsCfg, pw)
+	}
+}
+
+// serveListenConn handles one accepted connection: TLS + password auth, then a
+// single role request that it dispatches. Each connection is one role; the
+// manager and the destination open separate connections.
+func serveListenConn(raw net.Conn, tlsCfg *tls.Config, password string) {
+	remote := raw.RemoteAddr().String()
+	defer raw.Close()
+
+	conn := tls.Server(raw, tlsCfg)
+	const handshakeTimeout = 30 * time.Second
+	_ = conn.SetDeadline(time.Now().Add(handshakeTimeout))
+	if err := conn.Handshake(); err != nil {
+		fmt.Fprintf(os.Stderr, "listen: %s: tls handshake: %v\n", remote, err)
+		return
+	}
+	fr, fw, err := serverAuth(conn, password, handshakeTimeout)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "listen: %s: %v\n", remote, err)
+		return
+	}
+	_ = conn.SetDeadline(time.Time{}) // transfers set their own pacing
+
+	req, err := readCtrlFrame(fr)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "listen: %s: read role request: %v\n", remote, err)
+		return
+	}
+	if req.Type != msgRole {
+		_ = fw.WriteJSON(CtrlMsg{Type: msgRoleErr, Message: fmt.Sprintf("expected a role request, got %q", req.Type)})
+		return
+	}
+	if err := dispatchRole(req, fr, fw, remote); err != nil {
+		fmt.Fprintf(os.Stderr, "listen: %s: role %s: %v\n", remote, req.Role, err)
+	}
+}
+
+func dispatchRole(req CtrlMsg, fr *FrameReader, fw *FrameWriter, remote string) error {
+	switch req.Role {
+	case roleControl:
+		fmt.Fprintf(os.Stderr, "listen: %s: control session\n", remote)
+		return runControlAgent(fr, fw)
+	case roleSourceStream:
+		fmt.Fprintf(os.Stderr, "listen: %s: source-stream %s\n", remote, req.Path)
+		return runSourceStreamRole(fr, fw, req.Path, req.Base, req.Size, req.BlockSize)
+	case roleSink:
+		fmt.Fprintf(os.Stderr, "listen: %s: sink %s\n", remote, req.Path)
+		return runSinkRole(fr, fw, req.Path, req.Base, req.Size, req.BlockSize)
+	default:
+		_ = fw.WriteJSON(CtrlMsg{Type: msgRoleErr, Message: fmt.Sprintf("unknown role %q", req.Role)})
+		return fmt.Errorf("unknown role %q", req.Role)
+	}
+}
+
+// resolvePassword picks the shared password from, in order: --password-file, an
+// inline --password, then the CLONETOOL_PASSWORD environment variable. A
+// trailing newline in a password file is stripped.
+func resolvePassword(inline, file string) (string, error) {
+	if file != "" {
+		b, err := os.ReadFile(file)
+		if err != nil {
+			return "", fmt.Errorf("read password file: %w", err)
+		}
+		return strings.TrimRight(string(b), "\r\n"), nil
+	}
+	if inline != "" {
+		return inline, nil
+	}
+	return os.Getenv("CLONETOOL_PASSWORD"), nil
+}

main.go

@@ -4,7 +4,6 @@ import (
 	"flag"
 	"fmt"
 	"os"
-	"runtime"
 )
 
 func main() {
@@ -17,10 +16,12 @@ func main() {
 	switch os.Args[1] {
 	case "sync":
 		err = cmdSync(os.Args[2:])
+	case "listen":
+		err = cmdListen(os.Args[2:])
 	case "agent":
 		err = cmdAgent(os.Args[2:])
 	case "version", "--version":
-		fmt.Printf("clonetool %s/%s\n", runtime.GOOS, runtime.GOARCH)
+		fmt.Println(versionString())
 		return
 	case "-h", "--help", "help":
 		usage()
@@ -40,13 +41,19 @@ func usage() {
 
 Usage:
   clonetool sync --source LOC --dest LOC [options]
+  clonetool listen --bind ADDR --password SECRET
   clonetool version
   clonetool agent --role {control|sink|source-stream} ...   (internal, spawned automatically)
 
-LOC is either a local path, or [user@]host:path for a path reached over SSH.
+LOC is one of:
+  a local path                (e.g. /dev/sda, ./image.bin, C:\data\disk.img, \\.\PhysicalDrive0)
+  [user@]host:path            a path reached over SSH
+  tcp://host:port/path        a clonetool started with "listen" on that host (no SSH needed)
+
 Source, destination, and the machine running "sync" (the manager) may all be
 different machines: the manager only orchestrates, it never reads or writes
-a single block itself.
+a single block itself. A tcp:// source is a host running "clonetool listen"
+(e.g. a Windows box with no SSH server); the destination pulls from it directly.
 
 clonetool keeps no state between runs: every sync re-reads and re-hashes both
 the source and the destination and transfers only the blocks that differ.
@@ -65,6 +72,14 @@ Options for sync:
   --remote-bin PATH       path to clonetool on remote hosts (default "clonetool")
   --manager-host HOST     address peers should use to reach this machine, when
                           source or dest has no host part (defaults to the local hostname)
+  --password SECRET       shared password for a tcp:// listen endpoint
+  --password-file PATH    read the listen-endpoint password from a file
+                          (or set CLONETOOL_PASSWORD)
+
+Options for listen:
+  --bind ADDR             address to listen on (default ":9000")
+  --password SECRET       shared password (or --password-file, or CLONETOOL_PASSWORD)
+  --password-file PATH    read the shared password from a file
 `)
 }
 
@@ -89,12 +104,19 @@ func cmdSync(args []string) error {
 	sshBin := fs.String("ssh", "ssh", "ssh binary")
 	remoteBin := fs.String("remote-bin", "clonetool", "clonetool path on remote hosts")
 	managerHost := fs.String("manager-host", "", "address peers use to reach this machine")
+	password := fs.String("password", "", "shared password for a tcp:// listen endpoint")
+	passwordFile := fs.String("password-file", "", "read the listen-endpoint password from a file")
 	var sshOpts stringSlice
 	fs.Var(&sshOpts, "ssh-opt", `extra "-o OPT" passed to ssh (repeatable)`)
 	if err := fs.Parse(args); err != nil {
 		return err
 	}
 
+	pw, err := resolvePassword(*password, *passwordFile)
+	if err != nil {
+		return err
+	}
+
 	if *source == "" || *dest == "" {
 		fs.Usage()
 		return fmt.Errorf("--source and --dest are required")
@@ -113,6 +135,7 @@ func cmdSync(args []string) error {
 		Job: *job, Source: *source, Dest: *dest, BlockSize: blockSize,
 		Yes: *yes, Sudo: *sudoMode, Deploy: *deploy, ConnectTimeoutSec: *connectTimeout,
 		SSHBin: *sshBin, SSHOpts: sshOpts, RemoteBin: *remoteBin, ManagerHost: *managerHost,
+		Password: pw,
 	})
 }
 

manager.go

@@ -28,6 +28,9 @@ type SyncConfig struct {
 	SSHOpts           []string
 	RemoteBin         string
 	ManagerHost       string
+	// Password authenticates connections to a listen ("tcp://…") endpoint.
+	// Required when the source or dest is such an endpoint; unused for ssh/local.
+	Password string
 }
 
 func runSync(cfg SyncConfig) error {
@@ -42,18 +45,22 @@ func runSync(cfg SyncConfig) error {
 	if err := checkNotSame(srcSpec, dstSpec); err != nil {
 		return err
 	}
+	if (srcSpec.IsNet() || dstSpec.IsNet()) && cfg.Password == "" {
+		return fmt.Errorf("a tcp:// listen endpoint needs a password (--password, --password-file, or CLONETOOL_PASSWORD)")
+	}
 
-	// Make sure each remote endpoint has a runnable clonetool, copying this
+	// Make sure each ssh endpoint has a runnable clonetool, copying this binary
-	// binary over if not (unless --deploy=false). A host that appears on
+	// over if not (unless --deploy=false). A host that appears on both sides is
-	// both sides is only probed once.
+	// only probed once. Listen ("tcp://…") endpoints are skipped: the user
+	// started clonetool there manually, so there is nothing to deploy.
 	srcRemoteBin, dstRemoteBin := cfg.RemoteBin, cfg.RemoteBin
-	if !srcSpec.IsLocal() {
+	if !srcSpec.IsLocal() && !srcSpec.IsNet() {
 		if srcRemoteBin, err = resolveRemoteBin(&cfg, srcSpec, "source"); err != nil {
 			return err
 		}
 	}
-	if !dstSpec.IsLocal() {
+	if !dstSpec.IsLocal() && !dstSpec.IsNet() {
-		if !srcSpec.IsLocal() && sameHost(srcSpec, dstSpec) {
+		if !srcSpec.IsLocal() && !srcSpec.IsNet() && sameHost(srcSpec, dstSpec) {
 			dstRemoteBin = srcRemoteBin
 		} else if dstRemoteBin, err = resolveRemoteBin(&cfg, dstSpec, "dest"); err != nil {
 			return err
@@ -90,15 +97,27 @@ func runSync(cfg SyncConfig) error {
 	cb := transferCallbacks{onProgress: pp.print}
 
 	bothLocal := srcSpec.IsLocal() && dstSpec.IsLocal()
-	srcHost, srcUser := resolveConnectHost(srcSpec, &cfg)
-	dstHost, dstUser := resolveConnectHost(dstSpec, &cfg)
 
-	pushReq := CtrlMsg{
+	pushReq := CtrlMsg{Path: srcSpec.Path, Size: targetSize, BlockSize: cfg.BlockSize}
-		Path: srcSpec.Path, Size: targetSize, BlockSize: cfg.BlockSize,
+	fillPeer(&pushReq, dstSpec, &cfg, dstRemoteBin, bothLocal, dstSudo)
-		PeerHost: dstHost, PeerUser: dstUser, PeerPath: dstSpec.Path, PeerLocal: bothLocal,
+	pullReq := CtrlMsg{Path: dstSpec.Path, Size: targetSize, BlockSize: cfg.BlockSize}
-		RemoteBin: dstRemoteBin, SSHBin: cfg.SSHBin, SSHOpts: cfg.SSHOpts, ConnectTimeoutSec: cfg.ConnectTimeoutSec,
+	fillPeer(&pullReq, srcSpec, &cfg, srcRemoteBin, bothLocal, srcSudo)
-		Sudo: dstSudo,
+
+	if srcSpec.IsNet() {
+		// A listening source only accepts connections; it can't be driven to
+		// dial out for a push. The destination pulls from it directly.
+		fmt.Fprintf(os.Stderr, "source is a listener; pulling %s <- %s ...\n", dstSpec, srcSpec)
+		ok, reason, err := dstCtrl.ConnectPull(pullReq, cb)
+		if err != nil {
+			return err
 		}
+		if !ok {
+			pp.finish()
+			return fmt.Errorf(
+				"could not pull from the listening source (%s); check that the destination host can reach %s:%d and that --password matches the listener",
+				reason, srcSpec.Host, srcSpec.Port)
+		}
+	} else {
 		if bothLocal {
 			fmt.Fprintf(os.Stderr, "both source and dest are local; syncing directly (no ssh) ...\n")
 		} else {
@@ -111,12 +130,6 @@ func runSync(cfg SyncConfig) error {
 		if !ok {
 			pp.finish()
 			fmt.Fprintf(os.Stderr, "push not possible (%s); trying pull %s <- %s ...\n", reason, dstSpec, srcSpec)
-		pullReq := CtrlMsg{
-			Path: dstSpec.Path, Size: targetSize, BlockSize: cfg.BlockSize,
-			PeerHost: srcHost, PeerUser: srcUser, PeerPath: srcSpec.Path, PeerLocal: bothLocal,
-			RemoteBin: srcRemoteBin, SSHBin: cfg.SSHBin, SSHOpts: cfg.SSHOpts, ConnectTimeoutSec: cfg.ConnectTimeoutSec,
-			Sudo: srcSudo,
-		}
 			ok2, reason2, err2 := dstCtrl.ConnectPull(pullReq, cb)
 			if err2 != nil {
 				return err2
@@ -128,6 +141,7 @@ func runSync(cfg SyncConfig) error {
 					reason, reason2)
 			}
 		}
+	}
 
 	pp.finish()
 	label := cfg.Job
@@ -144,6 +158,20 @@ func runSync(cfg SyncConfig) error {
 // restart under sudo. The returned bool reports whether the agent (and any
 // peer helper it later spawns for this side) is running elevated.
 func bringUpController(spec Spec, tag string, cfg *SyncConfig, remoteBin, probePath string) (*Controller, PathInfo, bool, error) {
+	if spec.IsNet() {
+		// A listen endpoint is reached over TLS; sudo/self-deploy don't apply.
+		c, err := startNetworkController(spec, tag, cfg.Password, cfg.ConnectTimeoutSec)
+		if err != nil {
+			return nil, PathInfo{}, false, err
+		}
+		info, err := c.Stat(probePath)
+		if err != nil {
+			c.Close()
+			return nil, PathInfo{}, false, err
+		}
+		return c, info, false, nil
+	}
+
 	sudo := cfg.Sudo == "always" && canElevate()
 	c, err := startController(spec, tag, cfg, remoteBin, sudo)
 	if err != nil {
@@ -153,11 +181,6 @@ func bringUpController(spec Spec, tag string, cfg *SyncConfig, remoteBin, probeP
 	if err == nil {
 		return c, info, sudo, nil
 	}
-	if errors.Is(err, errNeedPriv) && !canElevate() {
-		c.Close()
-		return nil, PathInfo{}, sudo, fmt.Errorf(
-			"%w; on Windows, run clonetool from an elevated (Administrator) console to open a raw disk", err)
-	}
 	if errors.Is(err, errNeedPriv) && cfg.Sudo == "auto" && !sudo {
 		fmt.Fprintf(os.Stderr, "%s: permission denied on %s; retrying via sudo ...\n", tag, probePath)
 		c.Close()
@@ -189,6 +212,29 @@ func hostLabel(spec Spec) string {
 	return spec.Host
 }
 
+// fillPeer populates the peer-connection fields of a connect_push/connect_pull
+// request from the peer's Spec: a net (listen) peer is reached by TLS dial with
+// the shared password; an ssh/local peer keeps the ssh/local fields.
+func fillPeer(req *CtrlMsg, peer Spec, cfg *SyncConfig, peerRemoteBin string, bothLocal, sudo bool) {
+	req.PeerPath = peer.Path
+	req.PeerLocal = bothLocal
+	req.RemoteBin = peerRemoteBin
+	req.SSHBin = cfg.SSHBin
+	req.SSHOpts = cfg.SSHOpts
+	req.ConnectTimeoutSec = cfg.ConnectTimeoutSec
+	req.Sudo = sudo
+	if peer.IsNet() {
+		req.PeerNet = true
+		req.PeerHost = peer.Host
+		req.PeerPort = peer.Port
+		req.PeerPassword = cfg.Password
+		return
+	}
+	host, user := resolveConnectHost(peer, cfg)
+	req.PeerHost = host
+	req.PeerUser = user
+}
+
 func resolveConnectHost(spec Spec, cfg *SyncConfig) (host, user string) {
 	if !spec.IsLocal() {
 		return spec.Host, spec.User
@@ -234,28 +280,62 @@ func confirmShrink(dstSpec Spec, oldSize, newSize int64) bool {
 	return line == "y" || line == "yes"
 }
 
-// progressPrinter renders a single rewriting status line on stderr. Scan
+// rateTracker smooths a monotonically increasing block counter into a
-// and transfer now run concurrently, so it keeps the latest of each kind of
+// blocks/sec rate, sampling no more than a few times a second so a burst of
-// update (blocks hashed so far, blocks copied/skipped) and shows them on one
+// same-tick progress messages doesn't produce a noisy instantaneous rate.
-// line, with a rolling copy rate; the "scan H/M" segment disappears once the
+type rateTracker struct {
+	lastT   time.Time
+	lastVal int64
+	rate    float64
+}
+
+func (r *rateTracker) update(val int64) {
+	now := time.Now()
+	if r.lastT.IsZero() {
+		r.lastT, r.lastVal = now, val
+		return
+	}
+	dt := now.Sub(r.lastT).Seconds()
+	if dt < 0.4 {
+		return
+	}
+	inst := float64(val-r.lastVal) / dt
+	if r.rate == 0 {
+		r.rate = inst
+	} else {
+		r.rate = 0.6*r.rate + 0.4*inst
+	}
+	r.lastT, r.lastVal = now, val
+}
+
+func (r *rateTracker) mib(blockSize int64) float64 {
+	return r.rate * float64(blockSize) / (1024 * 1024)
+}
+
+// progressPrinter renders a single rewriting status line on stderr. Scan,
+// source read and destination write all run concurrently, each tracked by
+// its own phase-tagged progress message (see CtrlMsg.Phase), so it keeps the
+// latest of each and shows them on one line with their own rolling
+// throughput; the "scan H/M" segment and its read rate disappear once the
 // destination fingerprint is complete.
 type progressPrinter struct {
 	blockSize int64
-	start      time.Time
-	lastT      time.Time
-	lastCopied int64
-	rate       float64 // copied blocks/sec, smoothed
 	active    bool
 
 	hashed  int64
 	copied  int64
 	skipped int64
+	srcRead int64
+	written int64
 	total   int64
+
+	dstReadRate  rateTracker // destination scan/hash throughput
+	srcReadRate  rateTracker // source read/compare throughput
+	dstWriteRate rateTracker // destination write throughput
 }
 
 func newProgressPrinter(blockSize int64) *progressPrinter {
-	now := time.Now()
+	return &progressPrinter{blockSize: blockSize}
-	return &progressPrinter{blockSize: blockSize, start: now, lastT: now}
 }
 
 func (p *progressPrinter) print(m CtrlMsg) {
@@ -263,22 +343,22 @@ func (p *progressPrinter) print(m CtrlMsg) {
 	if m.TotalBlocks > 0 {
 		p.total = m.TotalBlocks
 	}
-	if m.Phase == "scan" {
+	switch m.Phase {
+	case "scan":
 		if m.Hashed > p.hashed {
 			p.hashed = m.Hashed
 		}
+		if p.total > 0 && p.hashed >= p.total {
+			p.dstReadRate.rate = 0 // scan finished; stop showing a frozen rate
 		} else {
-		p.copied, p.skipped = m.Copied, m.Skipped
+			p.dstReadRate.update(p.hashed)
-		now := time.Now()
-		if dt := now.Sub(p.lastT).Seconds(); dt >= 0.4 {
-			inst := float64(m.Copied-p.lastCopied) / dt
-			if p.rate == 0 {
-				p.rate = inst
-			} else {
-				p.rate = 0.6*p.rate + 0.4*inst
-			}
-			p.lastT, p.lastCopied = now, m.Copied
 		}
+	case "xfer":
+		p.copied, p.skipped, p.srcRead = m.Copied, m.Skipped, m.SrcRead
+		p.srcReadRate.update(p.srcRead)
+	case "write":
+		p.written = m.Written
+		p.dstWriteRate.update(p.written)
 	}
 
 	done := p.copied + p.skipped
@@ -286,15 +366,15 @@ func (p *progressPrinter) print(m CtrlMsg) {
 	if total <= 0 {
 		total = done
 	}
-	mib := p.rate * float64(p.blockSize) / (1024 * 1024)
 
 	scan := ""
 	if p.total > 0 && p.hashed < p.total {
 		scan = fmt.Sprintf("  scan %s/%s", formatCount(p.hashed), formatCount(p.total))
 	}
-	fmt.Fprintf(os.Stderr, "\r  syncing %s %s/%s  copied=%s skipped=%s%s  %6.1f MiB/s        ",
+	fmt.Fprintf(os.Stderr, "\r  syncing %s %s/%s  copied=%s skipped=%s%s  rd(src) %5.1f  rd(dst) %5.1f  wr(dst) %5.1f MiB/s        ",
 		progressBar(done, total), formatCount(done), formatCount(total),
-		formatCount(p.copied), formatCount(p.skipped), scan, mib)
+		formatCount(p.copied), formatCount(p.skipped), scan,
+		p.srcReadRate.mib(p.blockSize), p.dstReadRate.mib(p.blockSize), p.dstWriteRate.mib(p.blockSize))
 }
 
 // finish ends the current status line with a newline so following output

nettransport.go

@@ -0,0 +1,267 @@
+package main
+
+import (
+	"crypto/ecdsa"
+	"crypto/elliptic"
+	"crypto/hmac"
+	"crypto/pbkdf2"
+	"crypto/rand"
+	"crypto/sha256"
+	"crypto/tls"
+	"crypto/x509"
+	"crypto/x509/pkix"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"math/big"
+	"net"
+	"strconv"
+	"time"
+)
+
+// The network ("listen") transport is an alternative to ssh for reaching an
+// endpoint that has no ssh server — typically a Windows source started with
+// `clonetool listen`. It is:
+//
+//   - TLS 1.3 for confidentiality. The listener uses a fresh, self-signed,
+//     in-memory certificate generated at startup; the dialer does not verify
+//     it (there is no CA). Authentication is by shared password, below.
+//   - A password challenge-response *bound to the TLS session*. Both sides
+//     derive a key from the password with PBKDF2 and prove knowledge of it by
+//     HMAC'ing the connection's RFC 5705 exported keying material, which is
+//     unique to this specific TLS session. A man-in-the-middle terminating TLS
+//     sees a different exporter value on each leg, so it cannot produce a valid
+//     HMAC for either side even though the certificate is unverified. Both
+//     directions are checked, so each end authenticates the other.
+//
+// Once authenticated, the connection carries the ordinary frame protocol
+// (frame.go): first a msgRole request naming the endpoint to run, then that
+// role's traffic.
+
+const (
+	authExporterLabel = "EXPORTER-clonetool-auth-v1"
+	authKDFIterations = 200_000
+	authKeyLen        = 32
+	authExporterLen   = 32
+	authMsgType       = "auth"
+)
+
+// authKDFSalt is a fixed application salt. Per-connection uniqueness comes from
+// the TLS exporter that the HMAC is computed over, not from this salt; it only
+// domain-separates the derived key.
+var authKDFSalt = []byte("clonetool-listen-auth-salt-v1")
+
+func deriveAuthKey(password string) ([]byte, error) {
+	return pbkdf2.Key(sha256.New, password, authKDFSalt, authKDFIterations, authKeyLen)
+}
+
+// authTag is HMAC(key, side || exporter): "side" separates the client's proof
+// from the server's so the same value can't simply be reflected.
+func authTag(key []byte, side string, exporter []byte) []byte {
+	m := hmac.New(sha256.New, key)
+	m.Write([]byte(side))
+	m.Write(exporter)
+	return m.Sum(nil)
+}
+
+func exporterFor(cs tls.ConnectionState) ([]byte, error) {
+	return cs.ExportKeyingMaterial(authExporterLabel, nil, authExporterLen)
+}
+
+// newServerTLSConfig builds a TLS config with a fresh self-signed cert.
+func newServerTLSConfig() (*tls.Config, error) {
+	cert, err := selfSignedCert()
+	if err != nil {
+		return nil, err
+	}
+	return &tls.Config{
+		Certificates: []tls.Certificate{cert},
+		MinVersion:   tls.VersionTLS13,
+	}, nil
+}
+
+func selfSignedCert() (tls.Certificate, error) {
+	key, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+	if err != nil {
+		return tls.Certificate{}, err
+	}
+	serial, err := rand.Int(rand.Reader, new(big.Int).Lsh(big.NewInt(1), 128))
+	if err != nil {
+		return tls.Certificate{}, err
+	}
+	tmpl := x509.Certificate{
+		SerialNumber:          serial,
+		Subject:               pkix.Name{CommonName: "clonetool"},
+		NotBefore:             time.Now().Add(-time.Hour),
+		NotAfter:              time.Now().Add(24 * 365 * time.Hour),
+		KeyUsage:              x509.KeyUsageDigitalSignature | x509.KeyUsageKeyEncipherment,
+		ExtKeyUsage:           []x509.ExtKeyUsage{x509.ExtKeyUsageServerAuth},
+		BasicConstraintsValid: true,
+	}
+	der, err := x509.CreateCertificate(rand.Reader, &tmpl, &tmpl, &key.PublicKey, key)
+	if err != nil {
+		return tls.Certificate{}, err
+	}
+	return tls.Certificate{Certificate: [][]byte{der}, PrivateKey: key}, nil
+}
+
+// dialAuth dials addr, completes the TLS handshake and the mutual password
+// challenge, and returns the authenticated connection plus its frame
+// reader/writer. On any failure it closes the connection.
+func dialAuth(addr, password string, timeout time.Duration) (*tls.Conn, *FrameReader, *FrameWriter, error) {
+	d := net.Dialer{Timeout: timeout}
+	raw, err := d.Dial("tcp", addr)
+	if err != nil {
+		return nil, nil, nil, err
+	}
+	conn := tls.Client(raw, &tls.Config{InsecureSkipVerify: true, MinVersion: tls.VersionTLS13})
+	_ = conn.SetDeadline(time.Now().Add(timeout))
+	if err := conn.Handshake(); err != nil {
+		_ = conn.Close()
+		return nil, nil, nil, fmt.Errorf("tls handshake: %w", err)
+	}
+	key, err := deriveAuthKey(password)
+	if err != nil {
+		_ = conn.Close()
+		return nil, nil, nil, err
+	}
+	exporter, err := exporterFor(conn.ConnectionState())
+	if err != nil {
+		_ = conn.Close()
+		return nil, nil, nil, fmt.Errorf("tls keying material: %w", err)
+	}
+	fw := NewFrameWriter(conn)
+	fr := NewFrameReader(conn)
+
+	// Prove ourselves to the server, then verify the server back.
+	if err := fw.WriteJSON(CtrlMsg{Type: authMsgType, Message: hex.EncodeToString(authTag(key, "client", exporter))}); err != nil {
+		_ = conn.Close()
+		return nil, nil, nil, err
+	}
+	m, err := readCtrlFrame(fr)
+	if err != nil {
+		_ = conn.Close()
+		return nil, nil, nil, fmt.Errorf("authentication failed (bad password?): %w", err)
+	}
+	want := authTag(key, "server", exporter)
+	got, _ := hex.DecodeString(m.Message)
+	if m.Type != authMsgType || !hmac.Equal(got, want) {
+		_ = conn.Close()
+		return nil, nil, nil, fmt.Errorf("server authentication failed (password mismatch)")
+	}
+	_ = conn.SetDeadline(time.Time{}) // clear; per-op timeouts are handled by callers
+	return conn, fr, fw, nil
+}
+
+// serverAuth runs the server side of the password challenge on an
+// already-TLS-handshaken connection.
+func serverAuth(conn *tls.Conn, password string, timeout time.Duration) (*FrameReader, *FrameWriter, error) {
+	key, err := deriveAuthKey(password)
+	if err != nil {
+		return nil, nil, err
+	}
+	exporter, err := exporterFor(conn.ConnectionState())
+	if err != nil {
+		return nil, nil, fmt.Errorf("tls keying material: %w", err)
+	}
+	fw := NewFrameWriter(conn)
+	fr := NewFrameReader(conn)
+
+	_ = conn.SetReadDeadline(time.Now().Add(timeout))
+	m, err := readCtrlFrame(fr)
+	if err != nil {
+		return nil, nil, err
+	}
+	want := authTag(key, "client", exporter)
+	got, _ := hex.DecodeString(m.Message)
+	if m.Type != authMsgType || !hmac.Equal(got, want) {
+		return nil, nil, fmt.Errorf("client authentication failed (password mismatch)")
+	}
+	if err := fw.WriteJSON(CtrlMsg{Type: authMsgType, Message: hex.EncodeToString(authTag(key, "server", exporter))}); err != nil {
+		return nil, nil, err
+	}
+	_ = conn.SetReadDeadline(time.Time{})
+	return fr, fw, nil
+}
+
+// readCtrlFrame reads one frameCtrlJSON frame and decodes it into a CtrlMsg.
+func readCtrlFrame(fr *FrameReader) (CtrlMsg, error) {
+	typ, payload, err := fr.ReadFrame()
+	if err != nil {
+		return CtrlMsg{}, err
+	}
+	if typ != frameCtrlJSON {
+		return CtrlMsg{}, fmt.Errorf("expected control frame, got type %d", typ)
+	}
+	var m CtrlMsg
+	if err := json.Unmarshal(payload, &m); err != nil {
+		return CtrlMsg{}, err
+	}
+	return m, nil
+}
+
+// dialPeerLink connects to a listening clonetool over TLS, authenticates,
+// requests role for the given window, and waits for its READY handshake. It is
+// the net counterpart of spawnPeerLink (see agent.go).
+func dialPeerLink(req CtrlMsg, role string) (*peerLink, string) {
+	addr := net.JoinHostPort(req.PeerHost, strconv.Itoa(req.PeerPort))
+	timeout := time.Duration(req.ConnectTimeoutSec+2) * time.Second
+	conn, fr, fw, err := dialAuth(addr, req.PeerPassword, timeout)
+	if err != nil {
+		return nil, err.Error()
+	}
+	if err := fw.WriteJSON(CtrlMsg{
+		Type: msgRole, Role: role,
+		Path: req.PeerPath, Base: req.PeerBase, Size: req.Size, BlockSize: req.BlockSize,
+	}); err != nil {
+		_ = conn.Close()
+		return nil, err.Error()
+	}
+	if err := waitPeerReady(fr, timeout); err != nil {
+		_ = conn.Close()
+		return nil, err.Error()
+	}
+	return &peerLink{
+		fw:        fw,
+		fr:        fr,
+		stderr:    func() string { return "" },
+		finish:    conn.Close,
+		kill:      func() { _ = conn.Close() },
+		closeSend: conn.Close,
+	}, ""
+}
+
+// waitPeerReady waits for a data role's frameReady, surfacing a msgRoleErr the
+// listener may send instead when it can't start the requested role.
+func waitPeerReady(fr *FrameReader, timeout time.Duration) error {
+	type result struct {
+		typ     frameType
+		payload []byte
+		err     error
+	}
+	ch := make(chan result, 1)
+	go func() {
+		typ, payload, err := fr.ReadFrame()
+		ch <- result{typ, payload, err}
+	}()
+	select {
+	case r := <-ch:
+		if r.err != nil {
+			return fmt.Errorf("handshake failed: %w", r.err)
+		}
+		switch r.typ {
+		case frameReady:
+			return nil
+		case frameCtrlJSON:
+			var m CtrlMsg
+			if json.Unmarshal(r.payload, &m) == nil && m.Type == msgRoleErr {
+				return fmt.Errorf("remote refused role: %s", m.Message)
+			}
+			return fmt.Errorf("handshake failed: unexpected control frame")
+		default:
+			return fmt.Errorf("handshake failed: unexpected frame type %d", r.typ)
+		}
+	case <-time.After(timeout):
+		return fmt.Errorf("handshake timed out after %s", timeout)
+	}
+}

spec.go

@@ -3,21 +3,32 @@ package main
 import (
 	"fmt"
 	"path/filepath"
+	"strconv"
 	"strings"
 )
 
-// Spec is a parsed source/dest location: [user@]host:path, or a bare local
+// Spec is a parsed source/dest location. It is one of:
-// path (Host == "").
+//   - a bare local path              (Host == "", Net == false)
+//   - an ssh endpoint [user@]host:path (Host != "", Net == false)
+//   - a listen endpoint tcp://host:port/path (Net == true) — a clonetool
+//     started in "listen" mode that this side connects to over TLS instead
+//     of ssh. Only the source is expected to be a listener today.
 type Spec struct {
 	Raw  string
 	User string
 	Host string
 	Path string
+	Net  bool // tcp:// listen endpoint (connect to a listening clonetool)
+	Port int  // listener port, when Net
 }
 
-func (s Spec) IsLocal() bool { return s.Host == "" }
+func (s Spec) IsLocal() bool { return s.Host == "" && !s.Net }
+func (s Spec) IsNet() bool   { return s.Net }
 
 func (s Spec) String() string {
+	if s.Net {
+		return fmt.Sprintf("tcp://%s:%d/%s", s.Host, s.Port, s.Path)
+	}
 	if s.IsLocal() {
 		return s.Path
 	}
@@ -27,14 +38,38 @@ func (s Spec) String() string {
 	return fmt.Sprintf("%s:%s", s.Host, s.Path)
 }
 
-// parseSpec parses "[user@]host:path" or a local "path". A leading "/",
+// looksLikeLocalPath reports whether raw is an ordinary local path that must
-// "./" or "../", or the absence of any colon, is treated as a local path so
+// never be mistaken for a "host:path" spec: a unix absolute/relative path, a
-// that ordinary absolute/relative paths are never mistaken for a host spec.
+// Windows drive path (C:\... or C:/...), or a Windows UNC / raw-device path
+// (\\server\share, \\.\PhysicalDrive0).
+func looksLikeLocalPath(raw string) bool {
+	if strings.HasPrefix(raw, "/") || strings.HasPrefix(raw, "./") || strings.HasPrefix(raw, "../") {
+		return true
+	}
+	if strings.HasPrefix(raw, `\\`) { // UNC or \\.\ device
+		return true
+	}
+	// Windows drive path: a single letter, a colon, then a separator.
+	if len(raw) >= 3 && raw[1] == ':' &&
+		((raw[0] >= 'A' && raw[0] <= 'Z') || (raw[0] >= 'a' && raw[0] <= 'z')) &&
+		(raw[2] == '\\' || raw[2] == '/') {
+		return true
+	}
+	return false
+}
+
+// parseSpec parses a location: "tcp://host:port/path" (a listen endpoint),
+// "[user@]host:path" (ssh), or a local path. Local paths (unix or Windows,
+// including drive letters and \\.\ devices) and anything without a colon are
+// treated as local so an ordinary path is never mistaken for a host spec.
 func parseSpec(raw string) (Spec, error) {
 	if raw == "" {
 		return Spec{}, fmt.Errorf("empty location")
 	}
-	if strings.HasPrefix(raw, "/") || strings.HasPrefix(raw, "./") || strings.HasPrefix(raw, "../") || !strings.Contains(raw, ":") {
+	if strings.HasPrefix(raw, "tcp://") {
+		return parseNetSpec(raw)
+	}
+	if looksLikeLocalPath(raw) || !strings.Contains(raw, ":") {
 		return Spec{Raw: raw, Path: raw}, nil
 	}
 	idx := strings.Index(raw, ":")
@@ -55,15 +90,47 @@ func parseSpec(raw string) (Spec, error) {
 	return Spec{Raw: raw, User: user, Host: host, Path: path}, nil
 }
 
+// parseNetSpec parses "tcp://host:port/path". Everything after the first "/"
+// that follows the host:port authority is taken literally as the path, so a
+// Windows path (C:\dir\file, \\.\PhysicalDrive0) survives unmangled.
+func parseNetSpec(raw string) (Spec, error) {
+	rest := strings.TrimPrefix(raw, "tcp://")
+	slash := strings.IndexByte(rest, '/')
+	if slash < 0 {
+		return Spec{}, fmt.Errorf("cannot parse %q (expected tcp://host:port/path)", raw)
+	}
+	authority := rest[:slash]
+	path := rest[slash+1:]
+	if path == "" {
+		return Spec{}, fmt.Errorf("cannot parse %q: empty path", raw)
+	}
+	colon := strings.LastIndexByte(authority, ':')
+	if colon <= 0 || colon == len(authority)-1 {
+		return Spec{}, fmt.Errorf("cannot parse %q (expected tcp://host:port/path)", raw)
+	}
+	host := authority[:colon]
+	port, err := strconv.Atoi(authority[colon+1:])
+	if err != nil || port <= 0 || port > 65535 {
+		return Spec{}, fmt.Errorf("cannot parse %q: invalid port %q", raw, authority[colon+1:])
+	}
+	return Spec{Raw: raw, Host: host, Port: port, Path: path, Net: true}, nil
+}
+
 // checkNotSame does a best-effort local check that source and dest don't
 // refer to the exact same path, to avoid an obviously destructive mistake.
 // It cannot resolve whether two different remote hostnames are actually the
 // same machine.
 func checkNotSame(src, dst Spec) error {
+	if src.IsNet() != dst.IsNet() {
+		return nil
+	}
+	if src.IsNet() && (!strings.EqualFold(src.Host, dst.Host) || src.Port != dst.Port) {
+		return nil
+	}
 	if src.IsLocal() != dst.IsLocal() {
 		return nil
 	}
-	if !src.IsLocal() && !strings.EqualFold(src.Host, dst.Host) {
+	if !src.IsLocal() && !src.IsNet() && !strings.EqualFold(src.Host, dst.Host) {
 		return nil
 	}
 	if !src.IsLocal() && src.User != dst.User {

syncside.go

@@ -81,12 +81,14 @@ type blockHash struct {
 	hash  [32]byte
 }
 
-// streamHashBlocks reads f in blockSize-byte blocks up to size, hashing each
+// streamHashBlocks reads f in blockSize-byte blocks over the window
-// one and handing it to onHash the moment it is computed (this is what lets
+// [base, base+size), hashing each one and handing it to onHash the moment it
-// the peer start comparing before the whole side has been scanned). It keeps
+// is computed (this is what lets the peer start comparing before the whole
-// no hash state of its own. onProgress, if non-nil, is called with
+// side has been scanned). It keeps no hash state of its own. onProgress, if
-// (blocksHashed, totalBlocks) before the first block and after each one.
+// non-nil, is called with (blocksHashed, totalBlocks) before the first block
-func streamHashBlocks(f *os.File, size, blockSize, align int64, onHash func(bh blockHash) error, onProgress func(done, total int64)) error {
+// and after each one. base is 0 for a whole-file sync; a non-zero base
+// fingerprints just a byte window within a larger handle.
+func streamHashBlocks(f *os.File, base, size, blockSize, align int64, onHash func(bh blockHash) error, onProgress func(done, total int64)) error {
 	if err := checkBlockAlign(blockSize, align); err != nil {
 		return err
 	}
@@ -96,7 +98,7 @@ func streamHashBlocks(f *os.File, size, blockSize, align int64, onHash func(bh b
 		onProgress(0, blockCount)
 	}
 	for i := int64(0); i < blockCount; i++ {
-		n, err := readBlockAt(f, buf, i*blockSize, size, align)
+		n, err := readBlockAt(f, buf, base+i*blockSize, base+size, align)
 		if err != nil {
 			return fmt.Errorf("hash %s at block %d: %w", f.Name(), i, err)
 		}
@@ -130,11 +132,17 @@ func scanProgressEmitter(out *FrameWriter) func(done, total int64) {
 // never read twice.
 type sourceLoopParams struct {
 	File      *os.File // already opened for reading by the caller
-	Size      int64
+	Base      int64    // byte offset the window starts at (0 for a whole-file sync)
+	Size      int64    // window length in bytes
 	BlockSize int64
 	Align     int64
 	Hashes    <-chan blockHash
 	Out       *FrameWriter
+	// OnRead is called immediately after a block has been read and hashed
+	// for comparison — whether it turns out to match or differ — so
+	// callers can track true source read throughput independent of the
+	// send/ack timing tracked by OnSkip/OnSend below.
+	OnRead func(index uint64)
 	OnSkip func(index uint64)
 	// OnSend is called for a differing block immediately before its DATA
 	// frame goes on the wire — never after. The push driver relies on this
@@ -151,12 +159,15 @@ func runSourceLoop(p sourceLoopParams) error {
 	f := p.File
 	buf := make([]byte, p.BlockSize)
 	for bh := range p.Hashes {
-		offset := int64(bh.index) * p.BlockSize
+		offset := p.Base + int64(bh.index)*p.BlockSize
-		n, err := readBlockAt(f, buf, offset, p.Size, p.Align)
+		n, err := readBlockAt(f, buf, offset, p.Base+p.Size, p.Align)
 		if err != nil {
 			return fmt.Errorf("read %s at block %d: %w", f.Name(), bh.index, err)
 		}
 		hash := sha256.Sum256(buf[:n])
+		if p.OnRead != nil {
+			p.OnRead(bh.index)
+		}
 		if hash == bh.hash {
 			if p.OnSkip != nil {
 				p.OnSkip(bh.index)
@@ -181,6 +192,7 @@ func runSourceLoop(p sourceLoopParams) error {
 // same process, before OnWritten is called).
 type destLoopParams struct {
 	File      *os.File // already opened for read/write by the caller
+	Base      int64    // byte offset the window starts at (0 for a whole-file sync)
 	BlockSize int64
 	Align     int64 // sector alignment for a raw disk handle, else 0/1
 	In        *FrameReader
@@ -189,6 +201,10 @@ type destLoopParams struct {
 	// pull mode, where the remote source-stream has no other channel back
 	// to the manager for progress updates). Sink-role callers leave it nil.
 	OnCtrlMsg func(CtrlMsg)
+	// OnWrite is called right after a block is durably written — the true
+	// destination write-throughput measurement point, as opposed to
+	// whatever "copied" count a driver derives from the send/ack protocol.
+	OnWrite func(index uint64)
 }
 
 func runDestLoop(p destLoopParams) error {
@@ -212,12 +228,15 @@ func runDestLoop(p destLoopParams) error {
 				}
 				return fmt.Errorf("block %d: hash mismatch after transfer", index)
 			}
-			if err := writeBlockAt(f, block, int64(index)*p.BlockSize, p.Align); err != nil {
+			if err := writeBlockAt(f, block, p.Base+int64(index)*p.BlockSize, p.Align); err != nil {
 				if p.AckOut != nil {
 					_ = p.AckOut.WriteFrame(frameErr, encodeErrFrame(index, err.Error()))
 				}
 				return fmt.Errorf("write block %d: %w", index, err)
 			}
+			if p.OnWrite != nil {
+				p.OnWrite(index)
+			}
 			if p.AckOut != nil {
 				if err := p.AckOut.WriteFrame(frameAck, encodeIndexFrame(index)); err != nil {
 					return err

util.go

@@ -2,9 +2,25 @@ package main
 
 import (
 	"fmt"
+	"os/exec"
+	"strings"
 	"sync"
 )
 
+// runCmd runs name with args, returning trimmed combined output.
+func runCmd(name string, args ...string) (string, error) {
+	out, err := exec.Command(name, args...).CombinedOutput()
+	return strings.TrimSpace(string(out)), err
+}
+
+// runCmdStdin is runCmd with a fixed stdin string (for tools that prompt).
+func runCmdStdin(stdin, name string, args ...string) (string, error) {
+	c := exec.Command(name, args...)
+	c.Stdin = strings.NewReader(stdin)
+	out, err := c.CombinedOutput()
+	return strings.TrimSpace(string(out)), err
+}
+
 // limitedBuffer keeps only the last maxLen bytes written to it — used to
 // capture a bounded tail of a subprocess's stderr for error messages
 // without risking unbounded memory growth from a noisy remote command.

version.go

@@ -0,0 +1,35 @@
+package main
+
+import (
+	"fmt"
+	"runtime"
+	"strings"
+)
+
+// buildTime is stamped at build time via -ldflags "-X main.buildTime=..."
+// (build.sh does this). A plain "go build" with no ldflags leaves it at
+// "dev" — still a meaningful, comparable value: two "dev" builds are only
+// ever the same binary bytes copied around (self-deploy), never two
+// independently built ones landing on the same value by chance.
+var buildTime = "dev"
+
+// versionString is what `clonetool version` prints, and what a remote
+// host's own `version` output is compared against (see remoteBuildTag) to
+// tell whether it's running the same build as this binary.
+func versionString() string {
+	return fmt.Sprintf("clonetool %s/%s build=%s", runtime.GOOS, runtime.GOARCH, buildTime)
+}
+
+// remoteBuildTag extracts the "build=..." tag from a clonetool `version`
+// line, ignoring its GOOS/GOARCH — those are expected to legitimately
+// differ across a cross-compiled deploy — and any surrounding ssh/shell
+// noise on other lines. Returns "" if the line isn't a clonetool version
+// line at all (e.g. "command not found").
+func remoteBuildTag(out string) string {
+	line := firstLine(out)
+	i := strings.Index(line, "build=")
+	if i < 0 {
+		return ""
+	}
+	return strings.TrimSpace(line[i+len("build="):])
+}