Add a validation for Advertising Data. Lengh should not be less than 1.

[buster84]

Without this validation, "panic: runtime error: slice bounds out of range" may happen. I added validation which ensure length should not be less than 1.

[moguriso]

I could make sure same issue is occurred.

adachi@raspberrypi:~ $ sudo ./main-linux-arm
2016/12/13 09:52:36 dev: hci0 up
2016/12/13 09:52:36 dev: hci0 down
2016/12/13 09:52:36 dev: hci0 opened
Scanning...

<---- waiting for just a second ---->

panic: runtime error: slice bounds out of range

goroutine 40 [running]:
panic(0x310c88, 0x1070a018)
/usr/lib/go/src/runtime/panic.go:481 +0x330
github.com/paypal/gatt.(*Advertisement).unmarshall(0x107fa050, 0x1070e24e, 0x11, 0x11, 0x0, 0x0)
./vendor/src/github.com/paypal/gatt/adv.go:99 +0xd14
github.com/paypal/gatt.(*device).Init.func3(0x107cc0c0)
./vendor/src/github.com/paypal/gatt/device_linux.go:97 +0x74
github.com/paypal/gatt/linux.(*HCI).handleAdvertisement(0x107625c0, 0x107e8243, 0xc, 0xc)
./vendor/src/github.com/paypal/gatt/linux/hci.go:262 +0x378
created by github.com/paypal/gatt/linux.(*HCI).handleLEMeta
./vendor/src/github.com/paypal/gatt/linux/hci.go:359 +0xc4

And confirmed this issue is fixed with this patch. I've checked into bluetooth kernel driver, probably b[] array is returned from the following function.

linux/net/bluetooth/hci_event.c

4899 static void hci_le_adv_report_evt(struct hci_dev *hdev, struct sk_buff *skb)
4900 {
4901         u8 num_reports = skb->data[0];
4902         void *ptr = &skb->data[1];
4903
4904         hci_dev_lock(hdev);
4905
4906         while (num_reports--) {
4907                 struct hci_ev_le_advertising_info *ev = ptr;

Minimum value of "num_reports" seems "1". Therefore this patch is looks good to me.
How do you think folks?