VMware fuzzy?!

Hey there,
for the moment I wanted to use the VMware server version 1.0.3 (from 4/26/07, Build 44356) for Linux. And it goes woooh with OpenSUSE 10.3 (x86_64):

Using 2.6.x kernel build system.
make: Entering directory `/tmp/vmware-config1/vmnet-only'
make -C /lib/modules/2.6.22.5-31-default/build/include/.. SUBDIRS=$PWD SRCROOT=$PWD/. modules
make[1]: Entering directory `/usr/src/linux-2.6.22.5-31'
 CC [M]  /tmp/vmware-config1/vmnet-only/driver.o
 CC [M]  /tmp/vmware-config1/vmnet-only/hub.o
 CC [M]  /tmp/vmware-config1/vmnet-only/userif.o
/tmp/vmware-config1/vmnet-only/userif.c: In function ‘VNetCopyDatagramToUser’:
/tmp/vmware-config1/vmnet-only/userif.c:630: error: ‘const struct sk_buff’ has no member named ‘h’
/tmp/vmware-config1/vmnet-only/userif.c:630: error: ‘const struct sk_buff’ has no member named ‘nh’
/tmp/vmware-config1/vmnet-only/userif.c:636: error: ‘const struct sk_buff’ has no member named ‘h’
make[2]: *** [/tmp/vmware-config1/vmnet-only/userif.o] Error 1
make[1]: *** [_module_/tmp/vmware-config1/vmnet-only] Error 2
make[1]: Leaving directory `/usr/src/linux-2.6.22.5-31'
make: *** [vmnet.ko] Error 2
make: Leaving directory `/tmp/vmware-config1/vmnet-only'
Unable to build the vmnet module.

Interesting, because in /usr/src/linux/include/linux/skbuff.h is sk_buff defined h as head, so you need to substitute "h" to "head" and "nh" to "network_header" :).

But the story goes on, this error happens in userif.c in you can go on bridge.c ...

By the way the working directory is /usr/lib/vmware/, the interesting archive is: /usr/lib/vmware/modules/source/vmnet.tar . This one you extract with "tar xf vmnet.tar", a directory vmnet-only is created, within the broken files (broken == using wrong skh definition). After the editing of the files (userif.c, bridge.c ...) pack it back with "tar cvf vmnet.tar vmnet-only".


Ok here is a patch which makes the VMware build (but this one has a lot of Warnings), I#m not sure if it hurts stability, so you're on you own.

diff -urN vmnet-only/bridge.c /usr/lib/vmware/modules/source/vmnet-only/bridge.c
--- vmnet-only/bridge.c 2007-04-13 05:24:10.000000000 +0200
+++ /usr/lib/vmware/modules/source/vmnet-only/bridge.c  2007-12-07 23:55:38.000000000 +0100
@@ -1083,12 +1083,12 @@
VNetBridgeComputeHeaderPos(struct sk_buff *skb) // IN: buffer to examine
{
   /* Maybe some kernel gets it right... */
-   if (skb->h.raw != skb->nh.raw) {
+   if (skb->head != skb->network_header) {
      return;
   }
   switch (be16_to_cpu(skb->protocol)) {
      case ETH_P_IP:
-         skb->h.raw = skb->nh.raw + (skb->nh.raw[0] & 0x0F) * 4;
+         *(skb->head) = skb->network_header + (skb->network_header & 0x0F) * 4;
        return;
      default:
         LOG(3, (KERN_DEBUG "Unknown EII protocol %04X: csum at %d\n",
@@ -1370,7 +1370,7 @@
#  endif

   if (bridge->smac) {
-      if (VNetCallSMACFunc(bridge->smac, &skb, skb->mac.raw,
+      if (VNetCallSMACFunc(bridge->smac, &skb, skb->mac_header,
                           SMAC_CheckPacketFromHost) !=
          PacketStatusForwardPacket) {
         LOG(4, (KERN_NOTICE "bridge-%s: packet dropped .\n",
@@ -1392,7 +1392,7 @@
#endif
#endif
  
-   skb_push(skb, skb->data - skb->mac.raw);
+   skb_push(skb, skb->data - skb->mac_header);
   LOG(3, (KERN_DEBUG "bridge-%s: receive %d\n",
          bridge->name, (int) skb->len));
   if (VNetBridgeIsGSO(skb)) {
diff -urN vmnet-only/userif.c /usr/lib/vmware/modules/source/vmnet-only/userif.c
--- vmnet-only/userif.c 2007-04-13 05:24:10.000000000 +0200
+++ /usr/lib/vmware/modules/source/vmnet-only/userif.c  2007-12-07 23:49:33.000000000 +0100
@@ -627,13 +627,13 @@
    */
   if (skb->pkt_type == PACKET_OUTGOING &&     /* Packet must be outgoing */
       skb->ip_summed == VM_CHECKSUM_PARTIAL &&        /* Without checksum */
-       skb->h.raw != skb->nh.raw &&            /* We must know where header is */
+       (sk_buff_data_t)(skb->head) != (sk_buff_data_t)(skb->network_header) &&         /* We must know where header is */
       skb->len == count) {                    /* No truncation may occur */
      size_t skl;
      int csum;
      u_int16_t csum16;
     
-      skl = skb->h.raw - skb->data;
+      skl = skb->head - skb->data;
      if (VNetCopyDatagram(skb, buf, skl)) {
        return -EFAULT;
      }
diff -urN vmnet-only/vnetInt.h /usr/lib/vmware/modules/source/vmnet-only/vnetInt.h
--- vmnet-only/vnetInt.h        2007-04-13 05:24:10.000000000 +0200
+++ /usr/lib/vmware/modules/source/vmnet-only/vnetInt.h 2007-12-07 23:40:48.000000000 +0100
@@ -25,8 +25,8 @@
#define DEV_QUEUE_XMIT(skb, dev, pri)  (                 \
    (skb)->dev = (dev),                                   \
    (skb)->priority = (pri),                              \
-    (skb)->mac.raw = (skb)->data,                         \
-    (skb)->nh.raw = (skb)->data + sizeof (struct ethhdr), \
+    (skb)->mac_header = (skb)->data,                         \
+    (skb)->network_header = (skb)->data + sizeof (struct ethhdr), \
    dev_queue_xmit(skb)                                   \
  )
#ifdef KERNEL_2_3_15

If you have better solution, let me know. I'm going to make it a better patch, which should fix some nasty warnings as

 CC [M]  /tmp/vmware-config13/vmnet-only/netif.o
 CC [M]  /tmp/vmware-config13/vmnet-only/bridge.o
/tmp/vmware-config13/vmnet-only/bridge.c: In function ‘VNetBridgeReceiveFromVNet’:
/tmp/vmware-config13/vmnet-only/bridge.c:568: warning: assignment makes integer from pointer without a cast
/tmp/vmware-config13/vmnet-only/bridge.c:568: warning: assignment makes integer from pointer without a cast
/tmp/vmware-config13/vmnet-only/bridge.c: In function ‘VNetBridgeComputeHeaderPos’:
/tmp/vmware-config13/vmnet-only/bridge.c:1086: warning: comparison between pointer and integer
/tmp/vmware-config13/vmnet-only/bridge.c: In function ‘VNetBridgeReceiveFromDev’:
/tmp/vmware-config13/vmnet-only/bridge.c:1374: warning: passing argument 3 of ‘VNetCallSMACFunc’ makes pointer from integer without a cast
/tmp/vmware-config13/vmnet-only/bridge.c:1395: warning: passing argument 2 of ‘skb_push’ makes integer from pointer without a cast

That's all, hopefully VMware tests it software the next time :).