Skip to contentSkip to conversation

Bluetooth: btintel_pcie: validate packet_len before skb_put_data

PATCH Superseded Kiran K 3 participants 5 messages Sep 2 – Sep 3 Bluetooth drivers

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Kiran K

author
#0 lore raw

PATCH 1/2 Bluetooth: btintel_pcie: validate packet_len before skb_put_data

btintel_pcie_submit_rx_work() reads packet_len from rfh_hdr without checking if it exceeds the RX buffer size. An oversized packet_len can lead to an out-of-bounds read in skb_put_data().

Validate packet_len to ensure it is non-zero and does not exceed BTINTEL_PCIE_BUFFER_SIZE - sizeof(*rfh_hdr), logging an error when invalid.

Fixes c2b636b3f788 ("Bluetooth: btintel_pcie: Add support for PCIe transport")
---
 drivers/bluetooth/btintel_pcie.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/bluetooth/btintel_pcie.c +2 -1
index 30923eaabed7..1eabb0c8326f 100644
@@ -1954,7 +1954,8 @@ static int btintel_pcie_submit_rx_work(struct btintel_pcie_data *data, u8 status
 	rfh_hdr = buf;
 
 	len = rfh_hdr->packet_len;
-	if (len <= 0) {
+	if (len == 0 || len > BTINTEL_PCIE_BUFFER_SIZE - sizeof(*rfh_hdr)) {
+		bt_dev_err(data->hdev, "Invalid packet_len %d", len);
 		ret = -EINVAL;
 		goto resubmit;
 	}
Signature · 1 line
2.54.0
author
#1 lore raw

PATCH 2/2 Bluetooth: btintel_pcie: fix tx_handle bounds off-by-one

Valid indices into txq->urbd0s/tfds/bufs are 0..txq->count-1, so tfd_index == txq->count is already out of range. Change the guard in btintel_pcie_msix_tx_handle() from '> txq->count' to '>= txq->count'.

Fixes c2b636b3f788 ("Bluetooth: btintel_pcie: Add support for PCIe transport")
---
 drivers/bluetooth/btintel_pcie.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/bluetooth/btintel_pcie.c +1 -1
index 1eabb0c8326f..e2cd27fef0c4 100644
@@ -1423,7 +1423,7 @@ static void btintel_pcie_msix_tx_handle(struct btintel_pcie_data *data)
 
 		urbd0 = &txq->urbd0s[cr_tia];
 
-		if (urbd0->tfd_index > txq->count)
+		if (urbd0->tfd_index >= txq->count)
 			return;
 
 		cr_tia = (cr_tia + 1) % txq->count;
Signature · 1 line
2.54.0
#2 lore raw
In reply to Kiran K

This is automated email and please do not reply to this email!

Dear submitter,

Thank you for submitting the patches to the linux bluetooth mailing list. This is a CI test results with your patch series:
PW Link:https://patchwork.kernel.org/project/bluetooth/list/?series=1156173

---Test result---

Test Summary:

CheckPatch                    PASS      1.47 seconds
VerifyFixes                   PASS      0.14 seconds
VerifySignedoff               PASS      0.14 seconds
GitLint                       PASS      0.68 seconds
SubjectPrefix                 PASS      0.26 seconds
BuildKernel                   PASS      26.76 seconds
CheckAllWarning               PASS      28.91 seconds
CheckSparse                   PASS      27.68 seconds
BuildKernel32                 PASS      25.35 seconds
CheckKernelLLVM               SKIP      0.00 seconds
TestRunnerSetup               PASS      472.42 seconds
IncrementalBuild              PASS      26.78 seconds

Details
##############################
Test: CheckKernelLLVM - SKIP
Desc: Build kernel with LLVM + context analysis
Output:
Clang not found

https://github.com/bluez/bluetooth-next/pull/687

---
Regards,
Linux Bluetooth

#3 lore raw
In reply to Kiran K

Dear Kiran,

Thank you for your patch.

Am 02.09.26 um 15:18 schrieb Kiran K:

Quoted 7 lines

btintel_pcie_submit_rx_work() reads packet_len from rfh_hdr without checking if it exceeds the RX buffer size. An oversized packet_len can lead to an out-of-bounds read in skb_put_data().

Validate packet_len to ensure it is non-zero and does not exceed BTINTEL_PCIE_BUFFER_SIZE - sizeof(*rfh_hdr), logging an error when invalid.

Can this case be forced somehow? Please document how.

Did a tool find this case or did you experience it during testing/review? If a tool, please document it.

Quoted 11 lines
Fixes c2b636b3f788 ("Bluetooth: btintel_pcie: Add support for PCIe transport")
---
  drivers/bluetooth/btintel_pcie.c | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)
drivers/bluetooth/btintel_pcie.c +0 -0
index 30923eaabed7..1eabb0c8326f 100644

drivers/bluetooth/btintel_pcie.c static int btintel_pcie_submit_rx_work(struct btintel_pcie_data *data, u8 status · line 1958

Quoted 6 lines
drivers/bluetooth/btintel_pcie.c +2 -1
@@ -1954,7 +1954,8 @@ static int btintel_pcie_submit_rx_work(struct btintel_pcie_data *data, u8 status
  	rfh_hdr = buf;
 
  	len = rfh_hdr->packet_len;
-	if (len <= 0) {
+	if (len == 0 || len > BTINTEL_PCIE_BUFFER_SIZE - sizeof(*rfh_hdr)) {
+		bt_dev_err(data->hdev, "Invalid packet_len %d", len);

What about negative values for len as was checked before?

To have a more useful log, I’d also log the value of BTINTEL_PCIE_BUFFER_SIZE and sizeof(*rfh_hdr).

drivers/bluetooth/btintel_pcie.c static int btintel_pcie_submit_rx_work(struct btintel_pcie_data *data, u8 status · line 1961

Quoted 3 lines
drivers/bluetooth/btintel_pcie.c +0 -0
  		ret = -EINVAL;
  		goto resubmit;
  	}

Kind regards,

Paul

author
#4 lore raw

Hi Paul,

Thanks for your comments.

Quoted 13 lines

Subject: Re: [PATCH v1 1/2] Bluetooth: btintel_pcie: validate packet_len before skb_put_data

Dear Kiran,

Thank you for your patch.

Am 02.09.26 um 15:18 schrieb Kiran K:

Quoted 4 lines

btintel_pcie_submit_rx_work() reads packet_len from rfh_hdr without checking if it exceeds the RX buffer size. An oversized packet_len can lead to an out-of-bounds read in skb_put_data().

Quoted 5 lines
Quoted 3 lines

Validate packet_len to ensure it is non-zero and does not exceed BTINTEL_PCIE_BUFFER_SIZE - sizeof(*rfh_hdr), logging an error when invalid.

Can this case be forced somehow? Please document how.

Yes, this condition can be simulated in a test environment by either:
1. Using a customized firmware image configured to send a frame header with an invalid  packet_len  (e.g.,  0  or  > 4088 ).
2. Temporarily modifying  rfh_hdr->packet_len  in the driver code (or using a debug hook) right before  btintel_pcie_submit_rx_work()  is invoked.

Quoted 3 lines

Did a tool find this case or did you experience it during testing/review? If a tool, please document it.

Yes. This issue was reported by Mythos. I will document in commit message.

Quoted 15 lines

drivers/bluetooth/btintel_pcie.c static int btintel_pcie_submit_rx_work(struct

Quoted 14 lines
Fixes c2b636b3f788 ("Bluetooth: btintel_pcie: Add support for PCIe

transport")

---
  drivers/bluetooth/btintel_pcie.c | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)
drivers/bluetooth/btintel_pcie.c +0 -0
index 30923eaabed7..1eabb0c8326f 100644
@@ -1954,7 +1954,8 @@ static int btintel_pcie_submit_rx_work(struct

btintel_pcie_data *data, u8 status

Quoted 8 lines

drivers/bluetooth/btintel_pcie.c static int btintel_pcie_submit_rx_work(struct btintel_pcie_data *data, u8 status · line 1958

Quoted 6 lines
drivers/bluetooth/btintel_pcie.c +2 -1
  	rfh_hdr = buf;
 
  	len = rfh_hdr->packet_len;
-	if (len <= 0) {
+	if (len == 0 || len > BTINTEL_PCIE_BUFFER_SIZE - sizeof(*rfh_hdr)) {
+		bt_dev_err(data->hdev, "Invalid packet_len %d", len);

What about negative values for len as was checked before?

rfh_hdr->packet_len  is a 16-bit unsigned bitfield ( u64 packet_len:16 ), which takes values in the range  0  to  65535. When assigned to  int len , it is zero-extended and cannot produce a negative value ( len >= 0  is always true).

Quoted 3 lines

To have a more useful log, I’d also log the value of BTINTEL_PCIE_BUFFER_SIZE and sizeof(*rfh_hdr).

Ack.

Quoted 9 lines

drivers/bluetooth/btintel_pcie.c static int btintel_pcie_submit_rx_work(struct btintel_pcie_data *data, u8 status · line 1961

Quoted 4 lines
drivers/bluetooth/btintel_pcie.c +0 -0
  		ret = -EINVAL;
  		goto resubmit;
  	}

Kind regards,

Paul

Thanks,
Kiran