Bluetooth: Fix not checking advertisement bondaries
Subsystems (2)
Luiz Augusto von Dentz
maintainerPATCH v2 1/2 Bluetooth: Fix not checking advertisement bondaries
From: Luiz Augusto von Dentz <[email protected]>
When receiving advertisements check if the length is actually within the skb, this also make use of skb_pull to advance on the skb->data instead of a custom ptr that way skb->len shall always indicates how much data is remaining and can be used to perform checks if there is enough data to parse.
---
v2: Fixes rssi parsing.
net/bluetooth/hci_event.c | 73 ++++++++++++++++++++++++++++++--------- 1 file changed, 56 insertions(+), 17 deletions(-)
net/bluetooth/hci_event.c +56 -17
@@ -5599,24 +5599,41 @@ static void process_adv_report(struct hci_dev *hdev, u8 type, bdaddr_t *bdaddr,static void hci_le_adv_report_evt (struct hci_dev *hdev,struct sk_buff *skb) {u8 num_reports = skb->data[0 ]; -void *ptr = &skb->data[1 ];hci_dev_lock (hdev); +skb_pull (skb,sizeof (num_reports)); +while (num_reports--) { -struct hci_ev_le_advertising_info *ev = ptr; +struct hci_ev_le_advertising_info *ev;s8 rssi; -if (ev->length <=HCI_MAX_AD_LENGTH ) { - rssi = ev->data[ev->length]; -process_adv_report (hdev, ev->evt_type, &ev->bdaddr, - ev->bdaddr_type,NULL ,0 , rssi, - ev->data, ev->length, false); - }else { -bt_dev_err (hdev,"Dropping invalid advertising data" ); +if (skb->len <sizeof (*ev)) { +bt_dev_err (hdev,"Malformed advertising report" ); +break ; + } + + ev = (void *) skb->data; +skb_pull (skb,sizeof (*ev)); + +if (skb->len < ev->length || ev->length >HCI_MAX_AD_LENGTH ) { +bt_dev_err (hdev,"Malformed advertising data" ); +break ; } - ptr +=sizeof (*ev) + ev->length +1 ; +skb_pull (skb, ev->length); + +if (skb->len <sizeof (rssi)) { +bt_dev_err (hdev,"Malformed advertising rssi" ); +break ; + } + + rssi = skb->data[0 ]; +skb_pull (skb,sizeof (rssi)); + +process_adv_report (hdev, ev->evt_type, &ev->bdaddr, + ev->bdaddr_type,NULL ,0 , rssi, + ev->data, ev->length, false); }hci_dev_unlock (hdev);
@@ -5669,15 +5686,31 @@ static u8 ext_evt_type_to_legacy(struct hci_dev *hdev, u16 evt_type)static void hci_le_ext_adv_report_evt (struct hci_dev *hdev,struct sk_buff *skb) {u8 num_reports = skb->data[0 ]; -void *ptr = &skb->data[1 ];hci_dev_lock (hdev); +skb_pull (skb,sizeof (num_reports)); +while (num_reports--) { -struct hci_ev_le_ext_adv_report *ev = ptr; +struct hci_ev_le_ext_adv_report *ev;u8 legacy_evt_type;u16 evt_type; +if (skb->len <sizeof (*ev)) { +bt_dev_err (hdev,"Malformed ext advertising report" ); +break ; + } + + ev = (void *) skb->data; +skb_pull (skb,sizeof (*ev)); + +if (skb->len < ev->length || ev->length >HCI_MAX_AD_LENGTH ) { +bt_dev_err (hdev,"Malformed ext advertising data" ); +break ; + } + +skb_pull (skb, ev->length); + evt_type =__le16_to_cpu (ev->evt_type); legacy_evt_type =ext_evt_type_to_legacy (hdev, evt_type);if (legacy_evt_type !=LE_ADV_INVALID ) {
@@ -5687,7 +5720,6 @@ static void hci_le_ext_adv_report_evt(struct hci_dev *hdev, struct sk_buff *skb) !(evt_type & LE_EXT_ADV_LEGACY_PDU)); } - ptr +=sizeof (*ev) + ev->length; }hci_dev_unlock (hdev);
@@ -5873,19 +5905,26 @@ static void hci_le_direct_adv_report_evt(struct hci_dev *hdev, struct sk_buff *skb) {u8 num_reports = skb->data[0 ]; -void *ptr = &skb->data[1 ];hci_dev_lock (hdev); +skb_pull (skb,sizeof (num_reports)); +while (num_reports--) { -struct hci_ev_le_direct_adv_info *ev = ptr; +struct hci_ev_le_direct_adv_info *ev; + +if (skb->len <sizeof (*ev)) { +bt_dev_err (hdev,"Malformed direct advertising" ); +break ; + } + + ev = (void *) skb->data; +skb_pull (skb,sizeof (*ev));process_adv_report (hdev, ev->evt_type, &ev->bdaddr, ev->bdaddr_type, &ev->direct_addr, ev->direct_addr_type, ev->rssi,NULL ,0 , false); - - ptr +=sizeof (*ev); }hci_dev_unlock (hdev);
Signature · 1 line
2.26.2
Luiz Augusto von Dentz
maintainerPATCH v2 2/2 Bluetooth: A2MP: Fix not setting request ID
From: Luiz Augusto von Dentz <[email protected]>
This fixes not resetting of the request ID when sending A2MP_GETAMPASSOC_RSP.
--- net/bluetooth/a2mp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
net/bluetooth/a2mp.c +2 -1
@@ -381,10 +381,11 @@ static int a2mp_getampassoc_req(struct amp_mgr *mgr, struct sk_buff *skb, hdev =hci_dev_get (req->id);if (!hdev || hdev->amp_type ==AMP_TYPE_BREDR || tmp) {struct a2mp_amp_assoc_rsp rsp; - rsp.id = req->id;memset (&rsp,0 ,sizeof (rsp)); + rsp.id = req->id; +if (tmp) { rsp.status =A2MP_STATUS_COLLISION_OCCURED ;amp_mgr_put (tmp);
Signature · 1 line
2.26.2
Re: [PATCH v2 2/2] Bluetooth: A2MP: Fix not setting request ID
Hi Luiz,
Quoted 8 lines · Luiz Augusto von Dentz
This fixes not resetting of the request ID when sending
A2MP_GETAMPASSOC_RSP.Fixes Bluetooth: A2MP: Fix not initializing all membersSigned-off-by Luiz Augusto von Dentz <[email protected]>--- net/bluetooth/a2mp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
The fix itself looks fine, but I think the "Fixes: ..." line should contain the commit id followed by the commit summary within parentheses, at least based on how I can see this style of annotation used in other commits in the tree.
Johan
Luiz Augusto von Dentz
maintainerRe: [PATCH v2 2/2] Bluetooth: A2MP: Fix not setting request ID
Hi Johan,
Quoted 13 lines · Johan Hedberg
Hi Luiz,
Quoted 8 lines · Luiz Augusto von Dentz
This fixes not resetting of the request ID when sending
A2MP_GETAMPASSOC_RSP.Fixes Bluetooth: A2MP: Fix not initializing all membersSigned-off-by Luiz Augusto von Dentz <[email protected]>--- net/bluetooth/a2mp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
Quoted 4 lines
The fix itself looks fine, but I think the "Fixes: ..." line should contain the commit id followed by the commit summary within parentheses, at least based on how I can see this style of annotation used in other commits in the tree.
I did not add a commit id because it did not reached the Linus tree yet and we cannot use the one from bluetooth-next since that tree is rebased then the commit id might change so I thought the only way to do it is by commit message.
Signature · 1 line
Luiz Augusto von Dentz
Re: [PATCH v2 2/2] Bluetooth: A2MP: Fix not setting request ID
Hi Luiz,
Quoted 10 lines · Luiz Augusto von Dentz
Quoted 10 lines · Luiz Augusto von Dentz
Quoted 8 lines
This fixes not resetting of the request ID when sending
A2MP_GETAMPASSOC_RSP.Fixes Bluetooth: A2MP: Fix not initializing all membersSigned-off-by Luiz Augusto von Dentz <[email protected]>--- net/bluetooth/a2mp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
Quoted 7 lines
Quoted 4 lines
The fix itself looks fine, but I think the "Fixes: ..." line should contain the commit id followed by the commit summary within parentheses, at least based on how I can see this style of annotation used in other commits in the tree.
I did not add a commit id because it did not reached the Linus tree yet
Johan
Luiz Augusto von Dentz
maintainerRe: [PATCH v2 2/2] Bluetooth: A2MP: Fix not setting request ID
Hi Johan,
Quoted 19 lines · Johan Hedberg
Hi Luiz,
Quoted 15 lines · Luiz Augusto von Dentz
Quoted 14 lines · Luiz Augusto von Dentz
Quoted 8 lines
This fixes not resetting of the request ID when sending
A2MP_GETAMPASSOC_RSP.Fixes Bluetooth: A2MP: Fix not initializing all membersSigned-off-by Luiz Augusto von Dentz <[email protected]>--- net/bluetooth/a2mp.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)The fix itself looks fine, but I think the "Fixes: ..." line should contain the commit id followed by the commit summary within parentheses, at least based on how I can see this style of annotation used in other commits in the tree.
Quoted 5 lines
Quoted 2 lines
I did not add a commit id because it did not reached the Linus tree yet
Somehow I missed that, or git.kernel.org was playing tricks on me, Ive now updated the v3 to include it.
Quoted 1 line
Johan
Signature · 1 line
Luiz Augusto von Dentz
Quoted 9 lines · Luiz Augusto von Dentz
From: Luiz Augusto von Dentz <[email protected]>
When receiving advertisements check if the length is actually within the skb, this also make use of
skb_pullto advance on theskb->datainstead of a custom ptr that wayskb->lenshall always indicates how much data is remaining and can be used to perform checks if there is enough data to parse.
Quoted 7 lines
Fixes a2ec905d1e160a33b2e210e45ad30445ef26ce0e ("Bluetooth: fix kernel oops in store_pending_adv_report")Signed-off-by Luiz Augusto von Dentz <[email protected]>---
v2: Fixes rssi parsing.net/bluetooth/hci_event.c | 73 ++++++++++++++++++++++++++++++--------- 1 file changed, 56 insertions(+), 17 deletions(-)
---
I cherry-picked this to our 4.19 kernel and ran our LE Health tests on the Kohaku chromebook (AX201 controller). All tests are passing.
Hi Luiz,
Quoted 6 lines · Luiz Augusto von Dentz
When receiving advertisements check if the length is actually within the skb, this also make use of
skb_pullto advance on theskb->datainstead of a custom ptr that wayskb->lenshall always indicates how much data is remaining and can be used to perform checks if there is enough data to parse.
Quoted 7 lines
Fixes a2ec905d1e160a33b2e210e45ad30445ef26ce0e ("Bluetooth: fix kernel oops in store_pending_adv_report")Signed-off-by Luiz Augusto von Dentz <[email protected]>---
v2: Fixes rssi parsing.net/bluetooth/hci_event.c | 73 ++++++++++++++++++++++++++++++--------- 1 file changed, 56 insertions(+), 17 deletions(-)
Could we get the matching HCI logs for these corrupted events? It'd be good to include that in the commit message. Unless I misunderstood something, from what I can see from the changes the fields you are adding checks for are generated by the Bluetooth controller, i.e. only a buggy or broken Bluetooth controller would generate such events (meaning, this shouldn't be generally remotely exploitable), so it'd be good to know exactly which controllers generate such broken events.
Johan