panic error index out of range on read data by index

[mikepop7]

here is my config:
`
log, err := wal.Open("/tmp/wal/tidwall", &wal.Options{
SegmentSize: 128 * 1024 * 1024,
SegmentCacheSize: 1,
})

`

I start 2 go routine thread read segment data rand, and painc when read epos:
`
epos := s.epos[index-s.index]

`

it may caused by evicted itself when pushCache? LUR algorithm may could not put the segment specifying by segIdx to cache when this segment frequency less than others.

[tidwall]

Do you have the panic log?
I'd like to reproduce on my side.

[mikepop7]

Do you have the panic log? I'd like to reproduce on my side.

go.mod: github.com/tidwall/wal v1.1.8

func TestWriteData(t *testing.T) {
	log, err := wal.Open("/tmp/wal/tidwall-2", &wal.Options{
		SegmentSize:      16 * 1024 * 1024,
		SegmentCacheSize: 1,
	})
	fmt.Println(err)

	start := time.Now()
	data := make([]byte, 16*1024*1024)
	writeTimes := 1000
	for i := 0; i < writeTimes; i++ {
		log.Write(uint64(i), data)
	}
	fmt.Println("write data elapsed: ", time.Now().Sub(start))
}

func TestReadDataWithMultiThread(t *testing.T) {
	log, err := wal.Open("/tmp/wal/tidwall-2", &wal.Options{
		SegmentSize:      16 * 1024 * 1024,
		SegmentCacheSize: 1,
	})
	fmt.Println(err)

	wg := sync.WaitGroup{}
	wg.Add(1)

	go func() { //  thread A
		readBytes := 0
		startA := time.Now()
		bThreadStartFlag := false
		fmt.Println("A thread start", time.Now())

		for i := 1; i < 1000; i++ {
			read, err := log.Read(uint64(i))
			if err != nil {
				fmt.Println("A thread read error: ", err, ", index: ", i)
			}
			readBytes += len(read)
			if !bThreadStartFlag && readBytes > 5*1024*1024*1024 {
				bThreadStartFlag = true
				wg.Done()
			}
			if i%10000 == 0 {
				fmt.Println("A thread readBytes = ", ", i = ", i, time.Now())
			}
		}
		fmt.Println("A:", time.Now().Sub(startA), ", readBytes = ", readBytes)
	}()

	// B thread start after A read N bytes
	wg.Wait()
	fmt.Println("B thread start", time.Now())
	startB := time.Now()
	for i := 1; i < 1000; i++ {
		_, err := log.Read(uint64(i))
		if err != nil {
			fmt.Println("B thread read error: ", err, ", index: ", i)
		}
		if i%10000 == 0 {
			fmt.Println("B thread readBytes = ", ", i = ", i, time.Now())
		}
	}

	fmt.Println("B thread elapsed:", time.Now().Sub(startB))
}

TestReadDataWithMultiThread you could call multi times cause the bug to occur~

=== RUN   TestReadDataWithMultiThread
<nil>
A thread start 2025-09-08 09:51:08.3418314 +0800 CST m=+0.001043901
B thread start 2025-09-08 09:51:09.5059971 +0800 CST m=+1.165209601
panic: runtime error: index out of range [0] with length 0

goroutine 7 [running]:
github.com/tidwall/wal.(*Log).Read(0xc000146000, 0x283)
	/workspace/pkg/mod/github.com/tidwall/wal@v1.1.8/wal.go:635 +0x5ad
/my-test/wal_test_test.TestReadDataWithMultiThread.func1()
	/my-test/wal_test/tidwall_wal_test.go:44 +0x125
created by /my-test/wal_test_test.TestReadDataWithMultiThread in goroutine 6
	/my-test/wal_test/tidwall_wal_test.go:37 +0x11f