What happens when a deferred function panics while the goroutine is already panicking?

func main() {
	defer func() { fmt.Println("recovered:", recover()) }()
	defer func() { panic("second") }()
	panic("first")
}

Output: recovered: second. The new panic replaces the one in progress. Unwinding continues with the new value, and recover returns only the most recent panic value. The first value is lost to recover. If nothing recovers, the crash output lists the whole sequence:

panic: first
	panic: second

goroutine 1 [running]:
...

Practical consequences: cleanup code in a defer (closing files, unlocking, flushing) should not panic, because it hides the original cause. A recovery handler that logs recover() sees only the last panic. If cleanup can fail, return or log that error rather than panicking. Also remember that a panic in a deferred call still runs the remaining deferred calls in LIFO order.