fix: guard procCache.runLoop against panic leaving goroutine state corrupted

If runLoop panicked before closing firstReadyCh, callers would block until
context timeout, and s.running would remain true blocking all future restarts.
Defer now unconditionally resets s.running=false and closes firstReadyCh when
the panic happens before the first result is emitted.

Author: PhenixStar

pkg/wshrpc/wshremote/processviewer.go

@@ -89,17 +89,29 @@ func (s *procCacheState) requestAndWait(ctx context.Context) (*wshrpc.ProcessLis
 }
 
 func (s *procCacheState) runLoop(firstReadyCh chan struct{}) {
+	firstDone := false
 	defer func() {
 		panichandler.PanicHandler("procCache.runLoop", recover())
+		// Always clean up running state so future requests can restart the goroutine.
+		s.lock.Lock()
+		if !firstDone {
+			// Panic before first result: close channel so waiters unblock and get the
+			// "process list unavailable" error path via the nil cached check.
+			close(firstReadyCh)
+			s.ready = nil
+		}
+		s.running = false
+		s.lastCPUSamples = nil
+		s.lastCPUEpoch = 0
+		s.uidCache = nil
+		s.lock.Unlock()
 	}()
 
 	numCPU := runtime.NumCPU()
 	if numCPU < 1 {
 		numCPU = 1
 	}
 
-	firstDone := false
-
 	for {
 		iterStart := time.Now()