Skip to contentSkip to conversation

ath11k: fix kernel panic by freeing the msdu received with invalid length

PATCH In review Tamizh chelvam 2 participants 2 messages Nov 6, 2019 – Nov 22, 2019 Qualcomm Atheros ATH11K wireless driver

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ath11k: fix kernel panic by freeing the msdu received with invalid length

In certain scenario host receives the packets with invalid length which causes below kernel panic. Free up those msdus to avoid this kernel panic. And printed packet information for debugging purpose.

 2270.028121:   <6> task: ffffffc0008306d0 ti: ffffffc0008306d0 task.ti: ffffffc0008306d0
 2270.035247:   <2> PC is at skb_panic+0x40/0x44
 2270.042784:   <2> LR is at skb_panic+0x40/0x44
 2270.521775:   <2> [<ffffffc0004a06e0>] skb_panic+0x40/0x44
 2270.524039:   <2> [<ffffffc0004a1278>] skb_put+0x54/0x5c
 2270.529264:   <2> [<ffffffbffcc373a8>] ath11k_dp_process_rx_err+0x320/0x5b0 [ath11k]
 2270.533860:   <2> [<ffffffbffcc30b68>] ath11k_dp_service_srng+0x80/0x268 [ath11k]
 2270.541063:   <2> [<ffffffbffcc1d554>] ath11k_hal_rx_reo_ent_buf_paddr_get+0x200/0xb64 [ath11k]
 2270.547917:   <2> [<ffffffc0004b1f74>] net_rx_action+0xf8/0x274
 2270.556247:   <2> [<ffffffc000099df4>] __do_softirq+0x128/0x228
 2270.561625:   <2> [<ffffffc00009a130>] irq_exit+0x84/0xcc
 2270.567008:   <2> [<ffffffc0000cfb28>] __handle_domain_irq+0x8c/0xb0
 2270.571695:   <2> [<ffffffc000082484>] gic_handle_irq+0x6c/0xbc
---
 drivers/net/wireless/ath/ath11k/dp_rx.c | 43 +++++++++++++++++++++++++++++++++
 1 file changed, 43 insertions(+)
drivers/net/wireless/ath/ath11k/dp_rx.c +43 -0
index acad746..f2731cd 100644
@@ -1440,6 +1440,35 @@ static struct sk_buff *ath11k_dp_rx_get_msdu_last_buf(struct sk_buff_head *msdu_
 	return NULL;
 }
 
+static void ath11k_dp_rx_msdu_info(struct ath11k *ar,
+				   struct hal_rx_desc *rx_desc,
+				   struct ath11k_skb_rxcb *rxcb)
+{
+	enum hal_encrypt_type enctype;
+	bool is_decrypted;
+	bool mpdu_len_err;
+	u32 decap_format;
+	u8 *hdr_status;
+	u16 msdu_len;
+
+	hdr_status = ath11k_dp_rx_h_80211_hdr(rx_desc);
+	msdu_len = ath11k_dp_rx_h_msdu_start_msdu_len(rx_desc);
+	rxcb->is_first_msdu = ath11k_dp_rx_h_msdu_end_first_msdu(rx_desc);
+	rxcb->is_last_msdu = ath11k_dp_rx_h_msdu_end_last_msdu(rx_desc);
+
+	decap_format = ath11k_dp_rxdesc_get_decap_format(rx_desc);
+	mpdu_len_err = !!ath11k_dp_rxdesc_get_mpdulen_err(rx_desc);
+
+	is_decrypted = ath11k_dp_rx_h_attn_is_decrypted(rx_desc);
+	enctype = ath11k_dp_rx_h_mpdu_start_enctype(rx_desc);
+
+	ath11k_warn(ar->ab, "frame rx with invalid msdu len %u first msdu %d last msdu %d decap format %u mpdu_len_err %d decrypted %d encryption type %u\n",
+		    msdu_len, rxcb->is_first_msdu, rxcb->is_last_msdu,
+		    decap_format, mpdu_len_err, is_decrypted, enctype);
+	ath11k_dbg_dump(ar->ab, ATH11K_DBG_DATA, NULL, "", hdr_status,
+			sizeof(struct ieee80211_hdr));
+}
+
 static int ath11k_dp_rx_retrieve_amsdu(struct ath11k *ar,
 				       struct sk_buff_head *msdu_list,
 				       struct sk_buff_head *amsdu_list)
@@ -1492,6 +1521,12 @@ static int ath11k_dp_rx_retrieve_amsdu(struct ath11k *ar,
 		l3_pad_bytes = ath11k_dp_rx_h_msdu_end_l3pad(lrx_desc);
 
 		if (!rxcb->is_continuation) {
+			if ((msdu_len + HAL_RX_DESC_SIZE) > DP_RX_BUFFER_SIZE) {
+				ath11k_dbg_dump(ar->ab, ATH11K_DBG_DATA, NULL, "", rx_desc,
+						sizeof(struct hal_rx_desc));
+				ath11k_dp_rx_msdu_info(ar, rx_desc, rxcb);
+				goto free_out;
+			}
 			skb_put(msdu, HAL_RX_DESC_SIZE + l3_pad_bytes + msdu_len);
 			skb_pull(msdu, HAL_RX_DESC_SIZE + l3_pad_bytes);
 		} else {
@@ -2764,6 +2799,14 @@ static void ath11k_dp_rx_frag_h_mpdu(struct ath11k *ar,
 
 	rx_desc = (struct hal_rx_desc *)msdu->data;
 	msdu_len = ath11k_dp_rx_h_msdu_start_msdu_len(rx_desc);
+	if ((msdu_len + HAL_RX_DESC_SIZE) > DP_RX_BUFFER_SIZE) {
+		ath11k_dbg_dump(ar->ab, ATH11K_DBG_DATA, NULL, "", rx_desc,
+				sizeof(struct hal_rx_desc));
+		ath11k_dp_rx_msdu_info(ar, rx_desc, rxcb);
+		dev_kfree_skb_any(msdu);
+		goto exit;
+	}
+
 	skb_put(msdu, HAL_RX_DESC_SIZE + msdu_len);
 	skb_pull(msdu, HAL_RX_DESC_SIZE);
 
Signature · 7 lines
1.9.1


_______________________________________________
ath11k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath11k
#1 lore raw
Quoted 18 lines · Tamizh chelvam

In certain scenario host receives the packets with invalid length which causes below kernel panic. Free up those msdus to avoid this kernel panic. And printed packet information for debugging purpose.

 2270.028121:   <6> task: ffffffc0008306d0 ti: ffffffc0008306d0 task.ti: ffffffc0008306d0
 2270.035247:   <2> PC is at skb_panic+0x40/0x44
 2270.042784:   <2> LR is at skb_panic+0x40/0x44
 2270.521775:   <2> [<ffffffc0004a06e0>] skb_panic+0x40/0x44
 2270.524039:   <2> [<ffffffc0004a1278>] skb_put+0x54/0x5c
 2270.529264:   <2> [<ffffffbffcc373a8>] ath11k_dp_process_rx_err+0x320/0x5b0 [ath11k]
 2270.533860:   <2> [<ffffffbffcc30b68>] ath11k_dp_service_srng+0x80/0x268 [ath11k]
 2270.541063:   <2> [<ffffffbffcc1d554>] ath11k_hal_rx_reo_ent_buf_paddr_get+0x200/0xb64 [ath11k]
 2270.547917:   <2> [<ffffffc0004b1f74>] net_rx_action+0xf8/0x274
 2270.556247:   <2> [<ffffffc000099df4>] __do_softirq+0x128/0x228
 2270.561625:   <2> [<ffffffc00009a130>] irq_exit+0x84/0xcc
 2270.567008:   <2> [<ffffffc0000cfb28>] __handle_domain_irq+0x8c/0xb0
 2270.571695:   <2> [<ffffffc000082484>] gic_handle_irq+0x6c/0xbc
Quoted 1 line

Is your lastname uncapitalised on purpose? Looks odd.

Quoted 24 lines
drivers/net/wireless/ath/ath11k/dp_rx.c +18 -0
@@ -1440,6 +1440,35 @@ static struct sk_buff *ath11k_dp_rx_get_msdu_last_buf(struct sk_buff_head *msdu_
 	return NULL;
 }
 
+static void ath11k_dp_rx_msdu_info(struct ath11k *ar,
+				   struct hal_rx_desc *rx_desc,
+				   struct ath11k_skb_rxcb *rxcb)
+{
+	enum hal_encrypt_type enctype;
+	bool is_decrypted;
+	bool mpdu_len_err;
+	u32 decap_format;
+	u8 *hdr_status;
+	u16 msdu_len;
+
+	hdr_status = ath11k_dp_rx_h_80211_hdr(rx_desc);
+	msdu_len = ath11k_dp_rx_h_msdu_start_msdu_len(rx_desc);
+	rxcb->is_first_msdu = ath11k_dp_rx_h_msdu_end_first_msdu(rx_desc);
+	rxcb->is_last_msdu = ath11k_dp_rx_h_msdu_end_last_msdu(rx_desc);
+
+	decap_format = ath11k_dp_rxdesc_get_decap_format(rx_desc);
+	mpdu_len_err = !!ath11k_dp_rxdesc_get_mpdulen_err(rx_desc);

drivers/net/wireless/ath/ath11k/dp_rx.c static struct sk_buff *ath11k_dp_rx_get_msdu_last_buf(struct sk_buff_head *msdu_ · line 1470

Quoted 10 lines
drivers/net/wireless/ath/ath11k/dp_rx.c +10 -0
+
+	is_decrypted = ath11k_dp_rx_h_attn_is_decrypted(rx_desc);
+	enctype = ath11k_dp_rx_h_mpdu_start_enctype(rx_desc);
+
+	ath11k_warn(ar->ab, "frame rx with invalid msdu len %u first msdu %d last msdu %d decap format %u mpdu_len_err %d decrypted %d encryption type %u\n",
+		    msdu_len, rxcb->is_first_msdu, rxcb->is_last_msdu,
+		    decap_format, mpdu_len_err, is_decrypted, enctype);
+	ath11k_dbg_dump(ar->ab, ATH11K_DBG_DATA, NULL, "", hdr_status,
+			sizeof(struct ieee80211_hdr));
+}

This function feels quite excessive. The warning messages are supposed to be concise and printing 7 different values is far from that. So I would prefer to drop this function altogether.

drivers/net/wireless/ath/ath11k/dp_rx.c static int ath11k_dp_rx_retrieve_amsdu(struct ath11k *ar, · line 1529

Quoted 13 lines
drivers/net/wireless/ath/ath11k/dp_rx.c +6 -0
 static int ath11k_dp_rx_retrieve_amsdu(struct ath11k *ar,
 				       struct sk_buff_head *msdu_list,
 				       struct sk_buff_head *amsdu_list)
@@ -1492,6 +1521,12 @@ static int ath11k_dp_rx_retrieve_amsdu(struct ath11k *ar,
 		l3_pad_bytes = ath11k_dp_rx_h_msdu_end_l3pad(lrx_desc);
 
 		if (!rxcb->is_continuation) {
+			if ((msdu_len + HAL_RX_DESC_SIZE) > DP_RX_BUFFER_SIZE) {
+				ath11k_dbg_dump(ar->ab, ATH11K_DBG_DATA, NULL, "", rx_desc,
+						sizeof(struct hal_rx_desc));
+				ath11k_dp_rx_msdu_info(ar, rx_desc, rxcb);
+				goto free_out;
+			}

So instead I would just print a warning like this:

"dropping received amsdu frame with invalid msdu length %u"

And also call ath11k_dbg_dump() here.

Quoted 4 lines
drivers/net/wireless/ath/ath11k/dp_rx.c +0 -0
 			skb_put(msdu, HAL_RX_DESC_SIZE + l3_pad_bytes + msdu_len);
 			skb_pull(msdu, HAL_RX_DESC_SIZE + l3_pad_bytes);
 		} else {

drivers/net/wireless/ath/ath11k/dp_rx.c static void ath11k_dp_rx_frag_h_mpdu(struct ath11k *ar, · line 2808

Quoted 10 lines
drivers/net/wireless/ath/ath11k/dp_rx.c +7 -0
@@ -2764,6 +2799,14 @@ static void ath11k_dp_rx_frag_h_mpdu(struct ath11k *ar,
 
 	rx_desc = (struct hal_rx_desc *)msdu->data;
 	msdu_len = ath11k_dp_rx_h_msdu_start_msdu_len(rx_desc);
+	if ((msdu_len + HAL_RX_DESC_SIZE) > DP_RX_BUFFER_SIZE) {
+		ath11k_dbg_dump(ar->ab, ATH11K_DBG_DATA, NULL, "", rx_desc,
+				sizeof(struct hal_rx_desc));
+		ath11k_dp_rx_msdu_info(ar, rx_desc, rxcb);
+		dev_kfree_skb_any(msdu);
+		goto exit;
+	}

And the same here, but change the message a bit to make it unique to make it easier to find when reading logs:

"dropping received fragmented frame with invalid msdu length %u"

I just made up terms "amsdu" and "fragmented" from function names, please do come up with more appropriate terms if they are not correct.

Signature · 6 lines
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches

_______________________________________________
ath11k mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/ath11k