IB/rxe: Offload CRC calculation when possible

index b12dd9b5a89d..cfadd8d8593a 100644
@@ -64,6 +64,8 @@ static void rxe_cleanup(struct rxe_dev *rxe)
 	rxe_pool_cleanup(&rxe->mc_elem_pool);
 
 	rxe_cleanup_ports(rxe);
+
+	crypto_free_shash(rxe->tfm);
 }
 
 /* called when all references have been dropped */

index a696af81e4a5..ecdba2fce083 100644
@@ -50,6 +50,7 @@
 #include <rdma/ib_umem.h>
 #include <rdma/ib_cache.h>
 #include <rdma/ib_addr.h>
+#include <crypto/hash.h>
 
 #include "rxe_net.h"
 #include "rxe_opcode.h"

@@ -64,6 +65,25 @@
 
 #define RXE_ROCE_V2_SPORT		(0xc000)
 
+static inline u32 rxe_crc32(struct rxe_dev *rxe,
+			    u32 crc, void *next, size_t len)
+{
+	int err;
+
+	SHASH_DESC_ON_STACK(shash, rxe->tfm);
+
+	shash->tfm = rxe->tfm;
+	shash->flags = 0;
+	*(u32 *)shash_desc_ctx(shash) = crc;
+	err = crypto_shash_update(shash, next, len);
+	if (unlikely(err)) {
+		pr_warn_ratelimited("failed crc calculation, err: %d\n", err);
+		return crc32_le(crc, next, len);
+	}
+
+	return *(u32 *)shash_desc_ctx(shash);
+}
+
 int rxe_set_mtu(struct rxe_dev *rxe, unsigned int dev_mtu);
 
 int rxe_add(struct rxe_dev *rxe, unsigned int mtu);

index 413b56b23a06..39e0be31aab1 100644
@@ -87,10 +87,10 @@ u32 rxe_icrc_hdr(struct rxe_pkt_info *pkt, struct sk_buff *skb)
 	bth->qpn |= cpu_to_be32(~BTH_QPN_MASK);
 
 	length = hdr_size + RXE_BTH_BYTES;
-	crc = crc32_le(crc, pshdr, length);
+	crc = rxe_crc32(pkt->rxe, crc, pshdr, length);
 
 	/* And finish to compute the CRC on the remainder of the headers. */
-	crc = crc32_le(crc, pkt->hdr + RXE_BTH_BYTES,
-		       rxe_opcode[pkt->opcode].length - RXE_BTH_BYTES);
+	crc = rxe_crc32(pkt->rxe, crc, pkt->hdr + RXE_BTH_BYTES,
+			rxe_opcode[pkt->opcode].length - RXE_BTH_BYTES);
 	return crc;
 }

index 37eea7441ca4..154c3ee211ae 100644
@@ -370,7 +370,8 @@ int rxe_mem_copy(struct rxe_mem *mem, u64 iova, void *addr, int length,
 			((void *)(uintptr_t)iova) : addr;
 
 		if (crcp)
-			*crcp = crc32_le(*crcp, src, length);
+			crc = rxe_crc32(to_rdev(mem->pd->ibpd.device),
+					*crcp, src, length);
 
 		memcpy(dest, src, length);
 

@@ -403,7 +404,8 @@ int rxe_mem_copy(struct rxe_mem *mem, u64 iova, void *addr, int length,
 			bytes = length;
 
 		if (crcp)
-			crc = crc32_le(crc, src, bytes);
+			crc = rxe_crc32(to_rdev(mem->pd->ibpd.device),
+					crc, src, bytes);
 
 		memcpy(dest, src, bytes);
 

index 50886031096f..b1b268acc396 100644
@@ -387,8 +387,8 @@ int rxe_rcv(struct sk_buff *skb)
 	pack_icrc = be32_to_cpu(*icrcp);
 
 	calc_icrc = rxe_icrc_hdr(pkt, skb);
-	calc_icrc = crc32_le(calc_icrc, (u8 *)payload_addr(pkt),
-			     payload_size(pkt));
+	calc_icrc = rxe_crc32(rxe, calc_icrc, (u8 *)payload_addr(pkt),
+			      payload_size(pkt));
 	calc_icrc = (__force u32)cpu_to_be32(~calc_icrc);
 	if (unlikely(calc_icrc != pack_icrc)) {
 		if (skb->protocol == htons(ETH_P_IPV6))

index dbfde0dc6ff7..e9a6606f345f 100644
@@ -32,6 +32,7 @@
  */
 
 #include <linux/skbuff.h>
+#include <crypto/hash.h>
 
 #include "rxe.h"
 #include "rxe_loc.h"

@@ -483,8 +484,7 @@ static int fill_packet(struct rxe_qp *qp, struct rxe_send_wqe *wqe,
 		if (wqe->wr.send_flags & IB_SEND_INLINE) {
 			u8 *tmp = &wqe->dma.inline_data[wqe->dma.sge_offset];
 
-			crc = crc32_le(crc, tmp, paylen);
-
+			crc = rxe_crc32(rxe, crc, tmp, paylen);
 			memcpy(payload_addr(pkt), tmp, paylen);
 
 			wqe->dma.resid -= paylen;

index 5113e502f6f9..32526f644969 100644
@@ -1319,6 +1319,13 @@ int rxe_register_device(struct rxe_dev *rxe)
 	dev->attach_mcast = rxe_attach_mcast;
 	dev->detach_mcast = rxe_detach_mcast;
 
+	rxe->tfm = crypto_alloc_shash("crc32", 0, 0);
+	if (IS_ERR(rxe->tfm)) {
+		pr_err("failed to allocate crc algorithmi err:%ld",
+		       PTR_ERR(rxe->tfm));
+		return PTR_ERR(rxe->tfm);
+	}
+
 	err = ib_register_device(dev, NULL);
 	if (err) {
 		pr_warn("rxe_register_device failed, err = %d\n", err);

@@ -1339,6 +1346,8 @@ int rxe_register_device(struct rxe_dev *rxe)
 err2:
 	ib_unregister_device(dev);
 err1:
+	crypto_free_shash(rxe->tfm);
+
 	return err;
 }
 

index e100c500ae85..911939e1bb9d 100644
@@ -403,6 +403,7 @@ struct rxe_dev {
 
 	struct rxe_port		port;
 	struct list_head	list;
+	struct crypto_shash	*tfm;
 };
 
 static inline struct rxe_dev *to_rdev(struct ib_device *dev)

index d8610960630a..bf2c5324e107 100644
@@ -210,6 +210,39 @@ static struct dst_entry *rxe_find_route6(struct net_device *ndev,
 
 #endif
 
+static struct dst_entry *rxe_find_route(struct rxe_dev *rxe,
+					struct rxe_qp *qp,
+					struct rxe_av *av)
+{
+	struct dst_entry *dst = NULL;
+
+	if (qp_type(qp) == IB_QPT_RC)
+		dst = sk_dst_get(qp->sk->sk);
+
+	if (!dst || !(dst->obsolete && dst->ops->check(dst, 0))) {
+		if (dst)
+			dst_release(dst);
+
+		if (av->network_type == RDMA_NETWORK_IPV4) {
+			struct in_addr *saddr;
+			struct in_addr *daddr;
+
+			saddr = &av->sgid_addr._sockaddr_in.sin_addr;
+			daddr = &av->dgid_addr._sockaddr_in.sin_addr;
+			dst = rxe_find_route4(rxe->ndev, saddr, daddr);
+		} else if (av->network_type == RDMA_NETWORK_IPV6) {
+			struct in6_addr *saddr6;
+			struct in6_addr *daddr6;
+
+			saddr6 = &av->sgid_addr._sockaddr_in6.sin6_addr;
+			daddr6 = &av->dgid_addr._sockaddr_in6.sin6_addr;
+			dst = rxe_find_route6(rxe->ndev, saddr6, daddr6);
+		}
+	}
+
+	return dst;
+}
+
 static int rxe_udp_encap_recv(struct sock *sk, struct sk_buff *skb)
 {
 	struct udphdr *udph;

@@ -301,7 +334,7 @@ static void prepare_ipv4_hdr(struct dst_entry *dst, struct sk_buff *skb,
 	skb_scrub_packet(skb, xnet);
 
 	skb_clear_hash(skb);
-	skb_dst_set(skb, dst);
+	skb_dst_set(skb, dst_clone(dst));
 	memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
 
 	skb_push(skb, sizeof(struct iphdr));

@@ -349,13 +382,14 @@ static void prepare_ipv6_hdr(struct dst_entry *dst, struct sk_buff *skb,
 static int prepare4(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
 		    struct sk_buff *skb, struct rxe_av *av)
 {
+	struct rxe_qp *qp = pkt->qp;
 	struct dst_entry *dst;
 	bool xnet = false;
 	__be16 df = htons(IP_DF);
 	struct in_addr *saddr = &av->sgid_addr._sockaddr_in.sin_addr;
 	struct in_addr *daddr = &av->dgid_addr._sockaddr_in.sin_addr;
 
-	dst = rxe_find_route4(rxe->ndev, saddr, daddr);
+	dst = rxe_find_route(rxe, qp, av);
 	if (!dst) {
 		pr_err("Host not reachable\n");
 		return -EHOSTUNREACH;

@@ -369,17 +403,24 @@ static int prepare4(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
 
 	prepare_ipv4_hdr(dst, skb, saddr->s_addr, daddr->s_addr, IPPROTO_UDP,
 			 av->grh.traffic_class, av->grh.hop_limit, df, xnet);
+
+	if (qp_type(qp) == IB_QPT_RC)
+		sk_dst_set(qp->sk->sk, dst);
+	else
+		dst_release(dst);
+
 	return 0;
 }
 
 static int prepare6(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
 		    struct sk_buff *skb, struct rxe_av *av)
 {
-	struct dst_entry *dst;
+	struct rxe_qp *qp = pkt->qp;
+	struct dst_entry *dst = NULL;
 	struct in6_addr *saddr = &av->sgid_addr._sockaddr_in6.sin6_addr;
 	struct in6_addr *daddr = &av->dgid_addr._sockaddr_in6.sin6_addr;
 
-	dst = rxe_find_route6(rxe->ndev, saddr, daddr);
+	dst = rxe_find_route(rxe, qp, av);
 	if (!dst) {
 		pr_err("Host not reachable\n");
 		return -EHOSTUNREACH;

@@ -394,6 +435,12 @@ static int prepare6(struct rxe_dev *rxe, struct rxe_pkt_info *pkt,
 	prepare_ipv6_hdr(dst, skb, saddr, daddr, IPPROTO_UDP,
 			 av->grh.traffic_class,
 			 av->grh.hop_limit);
+
+	if (qp_type(qp) == IB_QPT_RC)
+		sk_dst_set(qp->sk->sk, dst);
+	else
+		dst_release(dst);
+
 	return 0;
 }
 

index f98a19e61a3d..aaaad3fb6849 100644
@@ -846,6 +846,14 @@ void rxe_qp_cleanup(struct rxe_pool_entry *arg)
 		qp->resp.mr = NULL;
 	}
 
+	if (qp_type(qp) == IB_QPT_RC) {
+		struct dst_entry *dst = NULL;
+
+		dst = sk_dst_get(qp->sk->sk);
+		if (dst)
+			dst_release(dst);
+	}
+
 	free_rd_atomic_resources(qp);
 
 	kernel_sock_shutdown(qp->sk, SHUT_RDWR);