d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix TSO support on VMware Fusion

Browse Source 
Apparently for VMware Fusion (and presumably VMware Workstation/Player
since the PR states TSO is broken there too, but I cannot test), the
TCP header pseudo checksum calculated should only include the protocol
(IPPROTO_TCP) value, not also the lengths as the stack does instead.

VMware ESXi seems to ignore whatever value is in the TCP header checksum,
and it is a bit surprising there is a different behavior between the
VMware products. And it is unfortunate that on ESXi we are forced to do
this extra bit of work.

PR:		kern/185849
MFC after:	3 days
This commit is contained in:
Bryan Venteicher 2014-06-09 02:45:53 +00:00
parent 75f4c88677
commit e41318fc7b
1 changed files with 45 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

54
sys/dev/vmware/vmxnet3/if_vmx.c
View File

@ -57,6 +57,8 @@ __FBSDID("$FreeBSD$");
#include <netinet/udp.h>
#include <netinet/tcp.h>
#include <machine/in_cksum.h>
#include <machine/bus.h>
#include <machine/resource.h>
#include <sys/bus.h>
@ -2604,6 +2606,12 @@ vmxnet3_txq_offload_ctx(struct vmxnet3_txqueue *txq, struct mbuf *m,
{
struct ether_vlan_header *evh;
int offset;
#if defined(INET)
struct ip *ip, iphdr;
#endif
#if defined(INET6)
struct ip6_hdr *ip6, ip6hdr;
#endif
evh = mtod(m, struct ether_vlan_header *);
if (evh->evl_encap_proto == htons(ETHERTYPE_VLAN)) {
@ -2617,8 +2625,7 @@ vmxnet3_txq_offload_ctx(struct vmxnet3_txqueue *txq, struct mbuf *m,
switch (*etype) {
#if defined(INET)
case ETHERTYPE_IP: {
struct ip *ip, iphdr;
case ETHERTYPE_IP:
if (__predict_false(m->m_len < offset + sizeof(struct ip))) {
m_copydata(m, offset, sizeof(struct ip),
(caddr_t) &iphdr);
@ -2628,10 +2635,16 @@ vmxnet3_txq_offload_ctx(struct vmxnet3_txqueue *txq, struct mbuf *m,
*proto = ip->ip_p;
*start = offset + (ip->ip_hl << 2);
break;
}
#endif
#if defined(INET6)
case ETHERTYPE_IPV6:
if (__predict_false(m->m_len <
offset + sizeof(struct ip6_hdr))) {
m_copydata(m, offset, sizeof(struct ip6_hdr),
(caddr_t) &ip6hdr);
ip6 = &ip6hdr;
} else
ip6 = mtodo(m, offset);
*proto = -1;
*start = ip6_lasthdr(m, offset, IPPROTO_IPV6, proto);
/* Assert the network stack sent us a valid packet. */
@ -2646,6 +2659,7 @@ vmxnet3_txq_offload_ctx(struct vmxnet3_txqueue *txq, struct mbuf *m,
if (m->m_pkthdr.csum_flags & CSUM_TSO) {
struct tcphdr *tcp, tcphdr;
uint16_t sum;
if (__predict_false(*proto != IPPROTO_TCP)) {
/* Likely failed to correctly parse the mbuf. */
@ -2654,16 +2668,38 @@ vmxnet3_txq_offload_ctx(struct vmxnet3_txqueue *txq, struct mbuf *m,
txq->vxtxq_stats.vmtxs_tso++;
switch (*etype) {
#if defined(INET)
case ETHERTYPE_IP:
sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr,
htons(IPPROTO_TCP));
break;
#endif
#if defined(INET6)
case ETHERTYPE_IPV6:
sum = in6_cksum_pseudo(ip6, 0, IPPROTO_TCP, 0);
break;
#endif
default:
sum = 0;
break;
}
if (m->m_len < *start + sizeof(struct tcphdr)) {
m_copyback(m, *start + offsetof(struct tcphdr, th_sum),
sizeof(uint16_t), (caddr_t) &sum);
m_copydata(m, *start, sizeof(struct tcphdr),
(caddr_t) &tcphdr);
tcp = &tcphdr;
} else {
tcp = mtodo(m, *start);
tcp->th_sum = sum;
}
/*
* For TSO, the size of the protocol header is also
* included in the descriptor header size.
*/
if (m->m_len < *start + sizeof(struct tcphdr)) {
m_copydata(m, offset, sizeof(struct tcphdr),
(caddr_t) &tcphdr);
tcp = &tcphdr;
} else
tcp = mtodo(m, *start);
*start += (tcp->th_off << 2);
} else
txq->vxtxq_stats.vmtxs_csum++;