Add bulkstat.go benchmarking program.

example/bulkstat.go

+package main
+
+// Given a file with filenames, this times how fast we can stat them
+// in parallel.  This is useful for benchmarking purposes.
+
+import (
+	"os"
+	"flag"
+	"time"
+	"fmt"
+	"encoding/line"
+)
+
+func main() {
+	flag.Parse()
+
+	filename := flag.Args()[0]
+	f, err := os.Open(filename, os.O_RDONLY, 0)
+	if err != nil {
+		panic("err"+err.String())
+	}
+
+	linelen := 1000
+	reader := line.NewReader(f, linelen)
+
+	files := make([]string, 0)
+	for {
+		l, _, err :=  reader.ReadLine()
+		if err != nil {
+			break
+		}
+		files = append(files, string(l))
+	}
+
+	parallel := 10
+	todo := make(chan string, len(files))
+	dts := make(chan int64, parallel)
+
+	for i := 0 ; i < parallel; i++ {
+		go func() {
+			for {
+				fn := <-todo
+				if fn == "" {
+					break
+				}
+
+				t := time.Nanoseconds()
+				os.Lstat(fn)
+				dts <- time.Nanoseconds() - t
+			}
+		}()
+	}
+
+	for _, v := range files {
+		todo <- v
+	}
+
+	total := 0.0
+	for i := 0 ; i < len(files); i++ {
+		total += float64(<-dts) * 1e-6
+	}
+
+	fmt.Println("Average stat time (ms):", total/float64(len(files)))
+}