Permit buiding kernels with options VIMAGE, restricted to only a single

active network stack instance.  Turning on options VIMAGE at compile
time yields the following changes relative to default kernel build:

1) V_ accessor macros for virtualized variables resolve to structure
fields via base pointers, instead of being resolved as fields in global
structs or plain global variables.  As an example, V_ifnet becomes:

    options VIMAGE:          ((struct vnet_net *) vnet_net)->_ifnet
    default build:           vnet_net_0._ifnet
    options VIMAGE_GLOBALS:  ifnet

2) INIT_VNET_* macros will declare and set up base pointers to be used
by V_ accessor macros, instead of resolving to whitespace:

    INIT_VNET_NET(ifp->if_vnet); becomes

    struct vnet_net *vnet_net = (ifp->if_vnet)->mod_data[VNET_MOD_NET];

3) Memory for vnet modules registered via vnet_mod_register() is now
allocated at run time in sys/kern/kern_vimage.c, instead of per vnet
module structs being declared as globals.  If required, vnet modules
can now request the framework to provide them with allocated bzeroed
memory by filling in the vmi_size field in their vmi_modinfo structures.

4) structs socket, ifnet, inpcbinfo, tcpcb and syncache_head are
extended to hold a pointer to the parent vnet.  options VIMAGE builds
will fill in those fields as required.

5) curvnet is introduced as a new global variable in options VIMAGE
builds, always pointing to the default and only struct vnet.

6) struct sysctl_oid has been extended with additional two fields to
store major and minor virtualization module identifiers, oid_v_subs and
oid_v_mod.  SYSCTL_V_* family of macros will fill in those fields
accordingly, and store the offset in the appropriate vnet container
struct in oid_arg1.
In sysctl handlers dealing with virtualized sysctls, the
SYSCTL_RESOLVE_V_ARG1() macro will compute the address of the target
variable and make it available in arg1 variable for further processing.

Unused fields in structs vnet_inet, vnet_inet6 and vnet_ipfw have
been deleted.

Reviewed by:	bz, rwatson
Approved by:	julian (mentor)

13 files changed, 39 insertions, 9 deletions

diff --git a/sys/netinet/in_pcb.c b/sys/netinet/in_pcb.c
index 9ff531a..dbc5ca8 100644
--- a/sys/netinet/in_pcb.c
+++ b/sys/netinet/in_pcb.c
@@ -126,7 +126,9 @@ sysctl_net_ipport_check(SYSCTL_HANDLER_ARGS)
 	INIT_VNET_INET(curvnet);
 	int error;
 
-	error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req);
+	SYSCTL_RESOLVE_V_ARG1();
+
+	error = sysctl_handle_int(oidp, arg1, arg2, req);
 	if (error == 0) {
 		RANGECHK(V_ipport_lowfirstauto, 1, IPPORT_RESERVED - 1);
 		RANGECHK(V_ipport_lowlastauto, 1, IPPORT_RESERVED - 1);
diff --git a/sys/netinet/in_pcb.h b/sys/netinet/in_pcb.h
index 82b4126..c86f1ab 100644
--- a/sys/netinet/in_pcb.h
+++ b/sys/netinet/in_pcb.h
@@ -224,6 +224,8 @@ struct inpcb {
 #define	in6p_icmp6filt	inp_depend6.inp6_icmp6filt
 #define	in6p_cksum	inp_depend6.inp6_cksum
 
+#define	inp_vnet	inp_pcbinfo->ipi_vnet
+
 /*
  * The range of the generation count, as used in this implementation, is 9e19.
  * We would have to create 300 billion connections per second for this number
@@ -301,8 +303,12 @@ struct inpcbinfo {
 	struct rwlock		 ipi_lock;
 
 	/*
-	 * vimage 1
-	 * general use 1
+	 * Pointer to network stack instance
+	 */
+	struct vnet		*ipi_vnet;
+
+	/*
+	 * general use 2
 	 */
 	void 			*ipi_pspare[2];
 };
diff --git a/sys/netinet/ip_divert.c b/sys/netinet/ip_divert.c
index f88a8fd..3bd3049 100644
--- a/sys/netinet/ip_divert.c
+++ b/sys/netinet/ip_divert.c
@@ -162,6 +162,9 @@ div_init(void)
 	INP_INFO_LOCK_INIT(&V_divcbinfo, "div");
 	LIST_INIT(&V_divcb);
 	V_divcbinfo.ipi_listhead = &V_divcb;
+#ifdef VIMAGE
+	V_divcbinfo.ipi_vnet = curvnet;
+#endif
 	/*
 	 * XXX We don't use the hash list for divert IP, but it's easier
 	 * to allocate a one entry hash list than it is to check all
diff --git a/sys/netinet/ip_fw.h b/sys/netinet/ip_fw.h
index fa37a73..cfc3089 100644
--- a/sys/netinet/ip_fw.h
+++ b/sys/netinet/ip_fw.h
@@ -695,7 +695,6 @@ struct vnet_ipfw {
 	int	_fw_deny_unknown_exthdrs;
 	int	_fw_verbose;
 	int	_verbose_limit;
-	int	_fw_debug;		/* actually unused */
 	int	_autoinc_step;
 	ipfw_dyn_rule **_ipfw_dyn_v;
 	uma_zone_t _ipfw_dyn_rule_zone;
@@ -740,7 +739,6 @@ extern struct vnet_ipfw vnet_ipfw_0;
 #define	V_fw_deny_unknown_exthdrs VNET_IPFW(fw_deny_unknown_exthdrs)
 #define	V_fw_verbose		VNET_IPFW(fw_verbose)
 #define	V_verbose_limit		VNET_IPFW(verbose_limit)
-#define	V_fw_debug		VNET_IPFW(fw_debug)
 #define	V_autoinc_step		VNET_IPFW(autoinc_step)
 #define	V_ipfw_dyn_v		VNET_IPFW(ipfw_dyn_v)
 #define	V_ipfw_dyn_rule_zone	VNET_IPFW(ipfw_dyn_rule_zone)
diff --git a/sys/netinet/ip_fw_pfil.c b/sys/netinet/ip_fw_pfil.c
index c2f10b1..3064cd9 100644
--- a/sys/netinet/ip_fw_pfil.c
+++ b/sys/netinet/ip_fw_pfil.c
@@ -95,6 +95,7 @@ int
 ipfw_check_in(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
     struct inpcb *inp)
 {
+	INIT_VNET_INET(curvnet);
 	struct ip_fw_args args;
 	struct ng_ipfw_tag *ng_tag;
 	struct m_tag *dn_tag;
@@ -224,6 +225,7 @@ int
 ipfw_check_out(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir,
     struct inpcb *inp)
 {
+	INIT_VNET_INET(curvnet);
 	struct ip_fw_args args;
 	struct ng_ipfw_tag *ng_tag;
 	struct m_tag *dn_tag;
diff --git a/sys/netinet/ip_input.c b/sys/netinet/ip_input.c
index a294d0a..7ee5a13 100644
--- a/sys/netinet/ip_input.c
+++ b/sys/netinet/ip_input.c
@@ -237,6 +237,7 @@ static void vnet_inet_register(void);
 static const vnet_modinfo_t vnet_inet_modinfo = {
 	.vmi_id		= VNET_MOD_INET,
 	.vmi_name	= "inet",
+	.vmi_size	= sizeof(struct vnet_inet)
 };
  
 static void vnet_inet_register()
diff --git a/sys/netinet/raw_ip.c b/sys/netinet/raw_ip.c
index 0775168..695a9cb 100644
--- a/sys/netinet/raw_ip.c
+++ b/sys/netinet/raw_ip.c
@@ -187,6 +187,9 @@ rip_init(void)
 
 	INP_INFO_LOCK_INIT(&V_ripcbinfo, "rip");
 	LIST_INIT(&V_ripcb);
+#ifdef VIMAGE
+	V_ripcbinfo.ipi_vnet = curvnet;
+#endif
 	V_ripcbinfo.ipi_listhead = &V_ripcb;
 	V_ripcbinfo.ipi_hashbase =
 	    hashinit(INP_PCBHASH_RAW_SIZE, M_PCB, &V_ripcbinfo.ipi_hashmask);
diff --git a/sys/netinet/tcp_subr.c b/sys/netinet/tcp_subr.c
index 9e92aab..c1dc4b3 100644
--- a/sys/netinet/tcp_subr.c
+++ b/sys/netinet/tcp_subr.c
@@ -359,6 +359,9 @@ tcp_init(void)
 
 	INP_INFO_LOCK_INIT(&V_tcbinfo, "tcp");
 	LIST_INIT(&V_tcb);
+#ifdef VIMAGE
+	V_tcbinfo.ipi_vnet = curvnet;
+#endif
 	V_tcbinfo.ipi_listhead = &V_tcb;
 	hashsize = TCBHASHSIZE;
 	TUNABLE_INT_FETCH("net.inet.tcp.tcbhashsize", &hashsize);
@@ -703,6 +706,9 @@ tcp_newtcpcb(struct inpcb *inp)
 	if (tm == NULL)
 		return (NULL);
 	tp = &tm->tcb;
+#ifdef VIMAGE
+	tp->t_vnet = inp->inp_vnet;
+#endif
 	tp->t_timers = &tm->tt;
 	/*	LIST_INIT(&tp->t_segq); */	/* XXX covered by M_ZERO */
 	tp->t_maxseg = tp->t_maxopd =
diff --git a/sys/netinet/tcp_syncache.c b/sys/netinet/tcp_syncache.c
index 18f3fb4..8e80842 100644
--- a/sys/netinet/tcp_syncache.c
+++ b/sys/netinet/tcp_syncache.c
@@ -259,6 +259,9 @@ syncache_init(void)
 
 	/* Initialize the hash buckets. */
 	for (i = 0; i < V_tcp_syncache.hashsize; i++) {
+#ifdef VIMAGE
+		V_tcp_syncache.hashbase[i].sch_vnet = curvnet;
+#endif
 		TAILQ_INIT(&V_tcp_syncache.hashbase[i].sch_bucket);
 		mtx_init(&V_tcp_syncache.hashbase[i].sch_mtx, "tcp_sc_head",
 			 NULL, MTX_DEF);
diff --git a/sys/netinet/tcp_syncache.h b/sys/netinet/tcp_syncache.h
index c367d33..e4e3fac 100644
--- a/sys/netinet/tcp_syncache.h
+++ b/sys/netinet/tcp_syncache.h
@@ -96,6 +96,7 @@ struct syncache {
 #define	SYNCOOKIE_LIFETIME	16	/* seconds */
 
 struct syncache_head {
+	struct vnet	*sch_vnet;
 	struct mtx	sch_mtx;
 	TAILQ_HEAD(sch_head, syncache)	sch_bucket;
 	struct callout	sch_timer;
diff --git a/sys/netinet/tcp_var.h b/sys/netinet/tcp_var.h
index 268db7c..5dc840e 100644
--- a/sys/netinet/tcp_var.h
+++ b/sys/netinet/tcp_var.h
@@ -35,6 +35,8 @@
 
 #include <netinet/tcp.h>
 
+struct vnet;
+
 /*
  * Kernel variables for tcp.
  */
@@ -106,6 +108,8 @@ struct tcpcb {
 	int	t_state;		/* state of this connection */
 	u_int	t_flags;
 
+	struct	vnet *t_vnet;		/* back pointer to parent vnet */
+
 	tcp_seq	snd_una;		/* send unacknowledged */
 	tcp_seq	snd_max;		/* highest sequence number sent;
 					 * used to recognize retransmits
@@ -186,8 +190,8 @@ struct tcpcb {
 	int	t_rttlow;		/* smallest observerved RTT */
 	u_int32_t	rfbuf_ts;	/* recv buffer autoscaling timestamp */
 	int	rfbuf_cnt;		/* recv buffer autoscaling byte count */
-	void	*t_pspare[3];		/* toe usrreqs / toepcb * / congestion algo / vimage / 1 general use */
-	struct toe_usrreqs *t_tu;       /* offload operations vector */
+	void	*t_pspare[3];		/* toe usrreqs / toepcb * / congestion algo / 1 general use */
+	struct toe_usrreqs *t_tu;	/* offload operations vector */
 	void	*t_toe;			/* TOE pcb pointer */
 	int	t_bytes_acked;		/* # bytes acked during current RTT */
 };
diff --git a/sys/netinet/udp_usrreq.c b/sys/netinet/udp_usrreq.c
index f2359eb..9aa83dd 100644
--- a/sys/netinet/udp_usrreq.c
+++ b/sys/netinet/udp_usrreq.c
@@ -179,6 +179,9 @@ udp_init(void)
 
 	INP_INFO_LOCK_INIT(&V_udbinfo, "udp");
 	LIST_INIT(&V_udb);
+#ifdef VIMAGE
+	V_udbinfo.ipi_vnet = curvnet;
+#endif
 	V_udbinfo.ipi_listhead = &V_udb;
 	V_udbinfo.ipi_hashbase = hashinit(UDBHASHSIZE, M_PCB,
 	    &V_udbinfo.ipi_hashmask);
diff --git a/sys/netinet/vinet.h b/sys/netinet/vinet.h
index 20a36c5..0057eff 100644
--- a/sys/netinet/vinet.h
+++ b/sys/netinet/vinet.h
@@ -54,7 +54,6 @@ struct vnet_inet {
 	struct	in_ifaddrhashhead *_in_ifaddrhashtbl;
 	struct	in_ifaddrhead _in_ifaddrhead;
 	u_long	_in_ifaddrhmask;
-	struct	in_multihead _in_multihead;	/* XXX unused */
 
 	int	_arpt_keep;
 	int	_arp_maxtries;
@@ -269,7 +268,6 @@ extern struct vnet_inet vnet_inet_0;
 #define	V_in_ifaddrhashtbl	VNET_INET(in_ifaddrhashtbl)
 #define	V_in_ifaddrhead		VNET_INET(in_ifaddrhead)
 #define	V_in_ifaddrhmask	VNET_INET(in_ifaddrhmask)
-#define	V_in_multihead		VNET_INET(in_multihead)
 #define	V_ip_checkinterface	VNET_INET(ip_checkinterface)
 #define	V_ip_defttl		VNET_INET(ip_defttl)
 #define	V_ip_do_randomid	VNET_INET(ip_do_randomid)