In the data setup code for doing offloads the

ip and tcp pointers were not reset after some
pullups. In practice this led to an NFS mount
failure when using UDP reported by Kevin Lo,
thanks Kevin. Fix from yongari, thank you!

1 files changed, 6 insertions, 1 deletions

diff --git a/sys/dev/e1000/if_em.c b/sys/dev/e1000/if_em.c
index 97d07db..1ce3956 100644
--- a/sys/dev/e1000/if_em.c
+++ b/sys/dev/e1000/if_em.c
@@ -93,7 +93,7 @@ int	em_display_debug_stats = 0;
 /*********************************************************************
  *  Driver version:
  *********************************************************************/
-char em_driver_version[] = "7.1.6";
+char em_driver_version[] = "7.1.7";
 
 /*********************************************************************
  *  PCI Device ID Table
@@ -1848,6 +1848,7 @@ em_xmit(struct tx_ring *txr, struct mbuf **m_headp)
 				*m_headp = NULL;
 				return (ENOBUFS);
 			}
+			ip = (struct ip *)(mtod(m_head, char *) + ip_off);
 			ip->ip_len = 0;
 			ip->ip_sum = 0;
 			/*
@@ -1856,6 +1857,7 @@ em_xmit(struct tx_ring *txr, struct mbuf **m_headp)
 			 * what hardware expect to see. This is adherence of
 			 * Microsoft's Large Send specification.
 			 */
+			tp = (struct tcphdr *)(mtod(m_head, char *) + poff);
 			tp->th_sum = in_pseudo(ip->ip_src.s_addr,
 			    ip->ip_dst.s_addr, htons(IPPROTO_TCP));
 		} else if (m_head->m_pkthdr.csum_flags & CSUM_TCP) {
@@ -1865,12 +1867,15 @@ em_xmit(struct tx_ring *txr, struct mbuf **m_headp)
 				*m_headp = NULL;
 				return (ENOBUFS);
 			}
+			ip = (struct ip *)(mtod(m_head, char *) + ip_off);
+			tp = (struct tcphdr *)(mtod(m_head, char *) + poff);
 		} else if (m_head->m_pkthdr.csum_flags & CSUM_UDP) {
 			m_head = m_pullup(m_head, poff + sizeof(struct udphdr));
 			if (m_head == NULL) {
 				*m_headp = NULL;
 				return (ENOBUFS);
 			}
+			ip = (struct ip *)(mtod(m_head, char *) + ip_off);
 		}
 		*m_headp = m_head;
 	}