fix(debug): target the traced thread and keep the kill from ever blocking

The Go runtime can migrate the debuggee's target-mode goroutine off the
process leader before PTRACE_TRACEME, which left the trace relation on a
thread the session never addressed: its stops starved the waits on the
leader, and a kill sequence that resumed nothing and then blocked in
Wait4 hung the whole package.  The debuggee now reports the traced thread
in the launch handshake and parks with a thread-directed stop, every
ptrace request and wait addresses that thread, a SIGURG arriving on a
single-step resumes it as a single-step again instead of letting the
tracee run uncontrolled, a resume rejected with ESRCH lifts a group-stop
with SIGCONT and retries once, and Kill resumes, kills and reaps through
non-blocking waits so it returns for a tracee in any state.

Assisted-by: GLM 5.3

debug/watchpoint_linux_amd64.go

@@ -36,7 +36,7 @@ const (
 // read, or the next hit on a different slot would still see this slot's bit
 // set and report this slot's address again.
 func archWatchpointAddr(s *Session, siAddr uint64) uint64 {
-	dr6, err := ptracePeekUser(s.pid, dr6Off)
+	dr6, err := ptracePeekUser(s.tid, dr6Off)
 	if err != nil {
 		return siAddr
 	}
@@ -46,10 +46,10 @@ func archWatchpointAddr(s *Session, siAddr uint64) uint64 {
 	}
 	// Best effort: the write-back only fails for a debuggee that died, in
 	// which case no further watchpoint can fire anyway.
-	_ = ptracePokeUser(s.pid, dr6Off, dr6&^0xF)
+	_ = ptracePokeUser(s.tid, dr6Off, dr6&^0xF)
 	for slot := range 4 {
 		if status&(1<<slot) != 0 {
-			addr, err := ptracePeekUser(s.pid, drOffset+uintptr(slot*8))
+			addr, err := ptracePeekUser(s.tid, drOffset+uintptr(slot*8))
 			if err == nil && addr != 0 {
 				return addr
 			}
@@ -112,11 +112,11 @@ func (s *Session) SetWatchpoint(slot int, addr uint64, typ WatchpointType, size
 		return fmt.Errorf("debug: watchpoint size must be 1, 2, 4, or 8")
 	}
 
-	if err := ptracePokeUser(s.pid, drOffset+uintptr(slot*8), addr); err != nil {
+	if err := ptracePokeUser(s.tid, drOffset+uintptr(slot*8), addr); err != nil {
 		return fmt.Errorf("debug: set DR%d: %w", slot, err)
 	}
 
-	dr7, err := ptracePeekUser(s.pid, dr7Off)
+	dr7, err := ptracePeekUser(s.tid, dr7Off)
 	if err != nil {
 		return fmt.Errorf("debug: read DR7: %w", err)
 	}
@@ -128,7 +128,7 @@ func (s *Session) SetWatchpoint(slot int, addr uint64, typ WatchpointType, size
 	mask := ^((uint64(1) << (2 * slot)) | (uint64(3) << (16 + 4*slot)) | (uint64(3) << (18 + 4*slot)))
 	dr7 = (dr7 & mask) | enableBit | rwBits | lenField
 
-	if err := ptracePokeUser(s.pid, dr7Off, dr7); err != nil {
+	if err := ptracePokeUser(s.tid, dr7Off, dr7); err != nil {
 		return fmt.Errorf("debug: set DR7: %w", err)
 	}
 	s.wpSlots[slot] = true
@@ -143,17 +143,17 @@ func (s *Session) ClearWatchpoint(slot int) error {
 	if !s.wpSlots[slot] {
 		return fmt.Errorf("debug: watchpoint slot %d is not in use", slot)
 	}
-	dr7, err := ptracePeekUser(s.pid, dr7Off)
+	dr7, err := ptracePeekUser(s.tid, dr7Off)
 	if err != nil {
 		return err
 	}
 	dr7 &^= uint64(1) << (2 * slot)
-	if err := ptracePokeUser(s.pid, dr7Off, dr7); err != nil {
+	if err := ptracePokeUser(s.tid, dr7Off, dr7); err != nil {
 		return err
 	}
 	// Zero the address register too: a stale address in a disabled slot
 	// turns any sticky DR6 bit into a misattributed watchpoint report.
-	if err := ptracePokeUser(s.pid, drOffset+uintptr(slot*8), 0); err != nil {
+	if err := ptracePokeUser(s.tid, drOffset+uintptr(slot*8), 0); err != nil {
 		return err
 	}
 	s.wpSlots[slot] = false