2021-11-24 20:16:42

by Gix, Brian

Subject: [PATCH v2] Bluetooth: refactor malicious adv data check

Check for out-of-bound read was being performed at the end of while
num_reports loop, and would fill journal with false positives. Added
check to beginning of loop processing so that it doesn't get checked
after ptr has been advanced.

Signed-off-by: Brian Gix <[email protected]>
---
net/bluetooth/hci_event.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/net/bluetooth/hci_event.c b/net/bluetooth/hci_event.c
index efc5458b1345..dee4ef22fc88 100644
--- a/net/bluetooth/hci_event.c
+++ b/net/bluetooth/hci_event.c
@@ -5920,6 +5920,11 @@ static void hci_le_adv_report_evt(struct hci_dev *hdev, struct sk_buff *skb)
struct hci_ev_le_advertising_info *ev = ptr;
s8 rssi;

+ if (ptr > (void *)skb_tail_pointer(skb) - sizeof(*ev)) {
+ bt_dev_err(hdev, "Malicious advertising data.");
+ break;
+ }
+
if (ev->length <= HCI_MAX_AD_LENGTH &&
ev->data + ev->length <= skb_tail_pointer(skb)) {
rssi = ev->data[ev->length];
@@ -5931,11 +5936,6 @@ static void hci_le_adv_report_evt(struct hci_dev *hdev, struct sk_buff *skb)
}

ptr += sizeof(*ev) + ev->length + 1;
-
- if (ptr > (void *) skb_tail_pointer(skb) - sizeof(*ev)) {
- bt_dev_err(hdev, "Malicious advertising data. Stopping processing");
- break;
- }
}

hci_dev_unlock(hdev);
--
2.31.1

2021-11-24 20:46:30

by Marcel Holtmann

Subject: Re: [PATCH v2] Bluetooth: refactor malicious adv data check

Hi Brian,

> Check for out-of-bound read was being performed at the end of while
> num_reports loop, and would fill journal with false positives. Added
> check to beginning of loop processing so that it doesn't get checked
> after ptr has been advanced.
>
> Signed-off-by: Brian Gix <[email protected]>
> ---
> net/bluetooth/hci_event.c | 10 +++++-----
> 1 file changed, 5 insertions(+), 5 deletions(-)

patch has been applied to bluetooth-next tree.

Regards

Marcel