netdev - [PATCH net 1/3] nfp: fix incorrectly set csum flag for nfd3 path

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu,  2 Mar 2023 10:58:28 +0100
From:   Simon Horman <simon.horman@...igine.com>
To:     David Miller <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>
Cc:     netdev@...r.kernel.org, oss-drivers@...igine.com,
        Huanhuan Wang <huanhuan.wang@...igine.com>,
        Louis Peens <louis.peens@...igine.com>,
        Simon Horman <simon.horman@...igine.com>
Subject: [PATCH net 1/3] nfp: fix incorrectly set csum flag for nfd3 path

From: Huanhuan Wang <huanhuan.wang@...igine.com>

The csum flag of IPsec packet are set repeatedly. Therefore, the csum
flag set of IPsec and non-IPsec packet need to be distinguished.

As the ipv6 header does not have a csum field, so l3-csum flag is not
required to be set for ipv6 case.

L4-csum flag include the tcp csum flag and udp csum flag, we shouldn't
set the udp and tcp csum flag at the same time for one packet, should
set l4-csum flag according to the transport layer is tcp or udp.

Fixes: 57f273adbcd4 ("nfp: add framework to support ipsec offloading")
Signed-off-by: Huanhuan Wang <huanhuan.wang@...igine.com>
Reviewed-by: Louis Peens <louis.peens@...igine.com>
Signed-off-by: Simon Horman <simon.horman@...igine.com>
---
 drivers/net/ethernet/netronome/nfp/nfd3/dp.c  |  7 +++---
 .../net/ethernet/netronome/nfp/nfd3/ipsec.c   | 25 +++++++++++++++++--
 2 files changed, 27 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfd3/dp.c b/drivers/net/ethernet/netronome/nfp/nfd3/dp.c
index 59fb0583cc08..0cc026b0aefd 100644
--- a/drivers/net/ethernet/netronome/nfp/nfd3/dp.c
+++ b/drivers/net/ethernet/netronome/nfp/nfd3/dp.c
@@ -324,14 +324,15 @@ netdev_tx_t nfp_nfd3_tx(struct sk_buff *skb, struct net_device *netdev)
 
 	/* Do not reorder - tso may adjust pkt cnt, vlan may override fields */
 	nfp_nfd3_tx_tso(r_vec, txbuf, txd, skb, md_bytes);
-	nfp_nfd3_tx_csum(dp, r_vec, txbuf, txd, skb);
+	if (ipsec)
+		nfp_nfd3_ipsec_tx(txd, skb);
+	else
+		nfp_nfd3_tx_csum(dp, r_vec, txbuf, txd, skb);
 	if (skb_vlan_tag_present(skb) && dp->ctrl & NFP_NET_CFG_CTRL_TXVLAN) {
 		txd->flags |= NFD3_DESC_TX_VLAN;
 		txd->vlan = cpu_to_le16(skb_vlan_tag_get(skb));
 	}
 
-	if (ipsec)
-		nfp_nfd3_ipsec_tx(txd, skb);
 	/* Gather DMA */
 	if (nr_frags > 0) {
 		__le64 second_half;
diff --git a/drivers/net/ethernet/netronome/nfp/nfd3/ipsec.c b/drivers/net/ethernet/netronome/nfp/nfd3/ipsec.c
index e90f8c975903..51087693072c 100644
--- a/drivers/net/ethernet/netronome/nfp/nfd3/ipsec.c
+++ b/drivers/net/ethernet/netronome/nfp/nfd3/ipsec.c
@@ -10,9 +10,30 @@
 void nfp_nfd3_ipsec_tx(struct nfp_nfd3_tx_desc *txd, struct sk_buff *skb)
 {
 	struct xfrm_state *x = xfrm_input_state(skb);
+	struct xfrm_offload *xo = xfrm_offload(skb);
+	struct iphdr *iph = ip_hdr(skb);
+	int l4_proto;
 
 	if (x->xso.dev && (x->xso.dev->features & NETIF_F_HW_ESP_TX_CSUM)) {
-		txd->flags |= NFD3_DESC_TX_CSUM | NFD3_DESC_TX_IP4_CSUM |
-			      NFD3_DESC_TX_TCP_CSUM | NFD3_DESC_TX_UDP_CSUM;
+		txd->flags |= NFD3_DESC_TX_CSUM;
+
+		if (iph->version == 4)
+			txd->flags |= NFD3_DESC_TX_IP4_CSUM;
+
+		if (x->props.mode == XFRM_MODE_TRANSPORT)
+			l4_proto = xo->proto;
+		else if (x->props.mode == XFRM_MODE_TUNNEL)
+			l4_proto = xo->inner_ipproto;
+		else
+			return;
+
+		switch (l4_proto) {
+		case IPPROTO_UDP:
+			txd->flags |= NFD3_DESC_TX_UDP_CSUM;
+			return;
+		case IPPROTO_TCP:
+			txd->flags |= NFD3_DESC_TX_TCP_CSUM;
+			return;
+		}
 	}
 }
-- 
2.30.2