svn commit: r191816 - in head/sys: conf dev/cxgb/ulp/tom kern net net80211 netgraph netinet netinet6 netipsec nfsclient sys

Marko Zec zec at FreeBSD.org
Tue May 5 10:56:14 UTC 2009


Author: zec
Date: Tue May  5 10:56:12 2009
New Revision: 191816
URL: http://svn.freebsd.org/changeset/base/191816

Log:
  Change the curvnet variable from a global const struct vnet *,
  previously always pointing to the default vnet context, to a
  dynamically changing thread-local one.  The currvnet context
  should be set on entry to networking code via CURVNET_SET() macros,
  and reverted to previous state via CURVNET_RESTORE().  Recursions
  on curvnet are permitted, though strongly discuouraged.
  
  This change should have no functional impact on nooptions VIMAGE
  kernel builds, where CURVNET_* macros expand to whitespace.
  
  The curthread->td_vnet (aka curvnet) variable's purpose is to be an
  indicator of the vnet context in which the current network-related
  operation takes place, in case we cannot deduce the current vnet
  context from any other source, such as by looking at mbuf's
  m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc.  Moreover, so
  far curvnet has turned out to be an invaluable consistency checking
  aid: it helps to catch cases when sockets, ifnets or any other
  vnet-aware structures may have leaked from one vnet to another.
  
  The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros
  was a result of an empirical iterative process, whith an aim to
  reduce recursions on CURVNET_SET() to a minimum, while still reducing
  the scope of CURVNET_SET() to networking only operations - the
  alternative would be calling CURVNET_SET() on each system call entry.
  In general, curvnet has to be set in three typicall cases: when
  processing socket-related requests from userspace or from within the
  kernel; when processing inbound traffic flowing from device drivers
  to upper layers of the networking stack, and when executing
  timer-driven networking functions.
  
  This change also introduces a DDB subcommand to show the list of all
  vnet instances.
  
  Approved by:	julian (mentor)

Modified:
  head/sys/conf/files
  head/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c
  head/sys/kern/init_main.c
  head/sys/kern/kern_fork.c
  head/sys/kern/kern_linker.c
  head/sys/kern/kern_vimage.c
  head/sys/kern/subr_pcpu.c
  head/sys/kern/sys_socket.c
  head/sys/kern/uipc_socket.c
  head/sys/kern/uipc_syscalls.c
  head/sys/kern/uipc_usrreq.c
  head/sys/net/bpf.c
  head/sys/net/if.c
  head/sys/net/if_clone.c
  head/sys/net/if_ethersubr.c
  head/sys/net/if_var.h
  head/sys/net/netisr.c
  head/sys/net/rtsock.c
  head/sys/net80211/ieee80211_freebsd.c
  head/sys/netgraph/netgraph.h
  head/sys/netinet/if_ether.c
  head/sys/netinet/igmp.c
  head/sys/netinet/in_rmx.c
  head/sys/netinet/ip_input.c
  head/sys/netinet/tcp_hostcache.c
  head/sys/netinet6/in6_rmx.c
  head/sys/netinet6/ip6_mroute.c
  head/sys/netinet6/mld6.c
  head/sys/netinet6/nd6.c
  head/sys/netinet6/nd6_nbr.c
  head/sys/netipsec/xform_tcp.c
  head/sys/nfsclient/nfs_vnops.c
  head/sys/sys/proc.h
  head/sys/sys/ucred.h
  head/sys/sys/vimage.h

Modified: head/sys/conf/files
==============================================================================
--- head/sys/conf/files	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/conf/files	Tue May  5 10:56:12 2009	(r191816)
@@ -1952,8 +1952,8 @@ kern/kern_time.c		standard
 kern/kern_timeout.c		standard
 kern/kern_umtx.c		standard
 kern/kern_uuid.c		standard
-kern/kern_xxx.c			standard
 kern/kern_vimage.c		standard
+kern/kern_xxx.c			standard
 kern/link_elf.c			standard
 kern/linker_if.m		standard
 kern/md4c.c			optional netsmb

Modified: head/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c
==============================================================================
--- head/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/dev/cxgb/ulp/tom/cxgb_cpl_io.c	Tue May  5 10:56:12 2009	(r191816)
@@ -1217,7 +1217,7 @@ install_offload_ops(struct socket *so)
  * receive window.
  */
 static __inline int
-select_rcv_wscale(int space)
+select_rcv_wscale(int space, struct vnet *vnet)
 {
 	INIT_VNET_INET(so->so_vnet);
 	int wscale = 0;
@@ -1326,7 +1326,7 @@ static inline unsigned int
 calc_opt0h(struct socket *so, int mtu_idx)
 {
 	struct tcpcb *tp = so_sototcpcb(so);
-	int wscale = select_rcv_wscale(tp->rcv_wnd);
+	int wscale = select_rcv_wscale(tp->rcv_wnd, so->so_vnet);
 	
 	return V_NAGLE((tp->t_flags & TF_NODELAY) == 0) |
 	    V_KEEP_ALIVE((so_options_get(so) & SO_KEEPALIVE) != 0) | F_TCAM_BYPASS |

Modified: head/sys/kern/init_main.c
==============================================================================
--- head/sys/kern/init_main.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/init_main.c	Tue May  5 10:56:12 2009	(r191816)
@@ -74,6 +74,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/malloc.h>
 #include <sys/conf.h>
 #include <sys/cpuset.h>
+#include <sys/vimage.h>
 
 #include <machine/cpu.h>
 
@@ -452,6 +453,9 @@ proc0_init(void *dummy __unused)
 	p->p_ucred->cr_uidinfo = uifind(0);
 	p->p_ucred->cr_ruidinfo = uifind(0);
 	p->p_ucred->cr_prison = NULL;	/* Don't jail it. */
+#ifdef VIMAGE
+	p->p_ucred->cr_vnet = LIST_FIRST(&vnet_head);
+#endif
 #ifdef AUDIT
 	audit_cred_kproc0(p->p_ucred);
 #endif

Modified: head/sys/kern/kern_fork.c
==============================================================================
--- head/sys/kern/kern_fork.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/kern_fork.c	Tue May  5 10:56:12 2009	(r191816)
@@ -68,6 +68,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/sdt.h>
 #include <sys/sx.h>
 #include <sys/signalvar.h>
+#include <sys/vimage.h>
 
 #include <security/audit/audit.h>
 #include <security/mac/mac_framework.h>
@@ -523,6 +524,11 @@ again:
 	td2->td_sigmask = td->td_sigmask;
 	td2->td_flags = TDF_INMEM;
 
+#ifdef VIMAGE
+	td2->td_vnet = NULL;
+	td2->td_vnet_lpush = NULL;
+#endif
+
 	/*
 	 * Duplicate sub-structures as needed.
 	 * Increase reference counts on shared objects.

Modified: head/sys/kern/kern_linker.c
==============================================================================
--- head/sys/kern/kern_linker.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/kern_linker.c	Tue May  5 10:56:12 2009	(r191816)
@@ -993,6 +993,12 @@ kern_kldload(struct thread *td, const ch
 		return (error);
 
 	/*
+	 * It's possible that kldloaded module will attach a new ifnet,
+	 * so vnet context must be set when this ocurs.
+	 */
+	CURVNET_SET(TD_TO_VNET(td));
+
+	/*
 	 * If file does not contain a qualified name or any dot in it
 	 * (kldname.ko, or kldname.ver.ko) treat it as an interface
 	 * name.
@@ -1019,6 +1025,7 @@ kern_kldload(struct thread *td, const ch
 		*fileid = lf->id;
 unlock:
 	KLD_UNLOCK();
+	CURVNET_RESTORE();
 	return (error);
 }
 
@@ -1056,6 +1063,7 @@ kern_kldunload(struct thread *td, int fi
 	if ((error = priv_check(td, PRIV_KLD_UNLOAD)) != 0)
 		return (error);
 
+	CURVNET_SET(TD_TO_VNET(td));
 	KLD_LOCK();
 	lf = linker_find_file_by_id(fileid);
 	if (lf) {
@@ -1092,6 +1100,7 @@ kern_kldunload(struct thread *td, int fi
 		PMC_CALL_HOOK(td, PMC_FN_KLD_UNLOAD, (void *) &pkm);
 #endif
 	KLD_UNLOCK();
+	CURVNET_RESTORE();
 	return (error);
 }
 

Modified: head/sys/kern/kern_vimage.c
==============================================================================
--- head/sys/kern/kern_vimage.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/kern_vimage.c	Tue May  5 10:56:12 2009	(r191816)
@@ -31,6 +31,8 @@
 #include <sys/cdefs.h>
 __FBSDID("$FreeBSD$");
 
+#include "opt_ddb.h"
+
 #include <sys/param.h>
 #include <sys/types.h>
 #include <sys/kernel.h>
@@ -38,6 +40,9 @@ __FBSDID("$FreeBSD$");
 #include <sys/malloc.h>
 #include <sys/systm.h>
 #include <sys/vimage.h>
+#ifdef DDB
+#include <ddb/ddb.h>
+#endif
 
 #ifndef VIMAGE_GLOBALS
 
@@ -51,8 +56,6 @@ static int vnet_mod_constructor(struct v
 static int vnet_mod_destructor(struct vnet_modlink *);
 
 #ifdef VIMAGE
-/* curvnet should be thread-local - this is only a temporary step. */
-struct vnet *curvnet;
 struct vnet_list_head vnet_head;
 #endif
 
@@ -183,7 +186,8 @@ vnet_mod_deregister_multi(const struct v
 	free(vml, M_VIMAGE);
 }
 
-static int vnet_mod_constructor(struct vnet_modlink *vml)
+static int
+vnet_mod_constructor(struct vnet_modlink *vml)
 {
 	const struct vnet_modinfo *vmi = vml->vml_modinfo;
 
@@ -303,7 +307,9 @@ vi_init(void *unused)
 	if (vnet == NULL)
 		panic("vi_alloc: malloc failed");
 	LIST_INSERT_HEAD(&vnet_head, vnet, vnet_le);
+	vnet->vnet_magic_n = VNET_MAGIC_N;
 
+	/* We MUST clear curvnet in vi_init_done before going SMP. */
 	curvnet = LIST_FIRST(&vnet_head);
 #endif
 }
@@ -313,6 +319,10 @@ vi_init_done(void *unused)
 {
 	struct vnet_modlink *vml_iter;
 
+#ifdef VIMAGE
+	curvnet = NULL;
+#endif
+
 	if (TAILQ_EMPTY(&vnet_modpending_head))
 		return;
 
@@ -327,5 +337,45 @@ vi_init_done(void *unused)
 
 SYSINIT(vimage, SI_SUB_VIMAGE, SI_ORDER_FIRST, vi_init, NULL);
 SYSINIT(vimage_done, SI_SUB_VIMAGE_DONE, SI_ORDER_FIRST, vi_init_done, NULL);
-
 #endif /* !VIMAGE_GLOBALS */
+
+#ifdef VIMAGE
+#ifdef DDB
+static void
+db_vnet_ptr(void *arg)
+{
+
+	if (arg)
+		db_printf(" %p", arg);
+	else
+#if SIZE_MAX == UINT32_MAX /* 32-bit arch */
+		db_printf("          0");
+#else /* 64-bit arch, most probaly... */
+		db_printf("                  0");
+#endif
+}
+
+DB_SHOW_COMMAND(vnets, db_show_vnets)
+{
+	VNET_ITERATOR_DECL(vnet_iter);
+
+#if SIZE_MAX == UINT32_MAX /* 32-bit arch */
+	db_printf("      vnet ifs socks");
+	db_printf("        net       inet      inet6      ipsec   netgraph\n");
+#else /* 64-bit arch, most probaly... */
+	db_printf("              vnet ifs socks");
+	db_printf("                net               inet              inet6              ipsec           netgraph\n");
+#endif
+	VNET_FOREACH(vnet_iter) {
+		db_printf("%p %3d %5d",
+		    vnet_iter, vnet_iter->ifccnt, vnet_iter->sockcnt);
+		db_vnet_ptr(vnet_iter->mod_data[VNET_MOD_NET]);
+		db_vnet_ptr(vnet_iter->mod_data[VNET_MOD_INET]);
+		db_vnet_ptr(vnet_iter->mod_data[VNET_MOD_INET6]);
+		db_vnet_ptr(vnet_iter->mod_data[VNET_MOD_IPSEC]);
+		db_vnet_ptr(vnet_iter->mod_data[VNET_MOD_NETGRAPH]);
+		db_printf("\n");
+	}
+}
+#endif
+#endif /* VIMAGE */

Modified: head/sys/kern/subr_pcpu.c
==============================================================================
--- head/sys/kern/subr_pcpu.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/subr_pcpu.c	Tue May  5 10:56:12 2009	(r191816)
@@ -135,6 +135,10 @@ show_pcpu(struct pcpu *pc)
 		db_printf("none\n");
 	db_show_mdpcpu(pc);
 		
+#ifdef VIMAGE
+	db_printf("curvnet      = %p\n", pc->pc_curthread->td_vnet);
+#endif
+
 #ifdef WITNESS
 	db_printf("spin locks held:\n");
 	witness_list_locks(&pc->pc_spinlocks);

Modified: head/sys/kern/sys_socket.c
==============================================================================
--- head/sys/kern/sys_socket.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/sys_socket.c	Tue May  5 10:56:12 2009	(r191816)
@@ -50,6 +50,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/stat.h>
 #include <sys/uio.h>
 #include <sys/ucred.h>
+#include <sys/vimage.h>
 
 #include <net/if.h>
 #include <net/route.h>
@@ -74,16 +75,19 @@ soo_read(struct file *fp, struct uio *ui
     int flags, struct thread *td)
 {
 	struct socket *so = fp->f_data;
-#ifdef MAC
 	int error;
 
+#ifdef MAC
 	SOCK_LOCK(so);
 	error = mac_socket_check_receive(active_cred, so);
 	SOCK_UNLOCK(so);
 	if (error)
 		return (error);
 #endif
-	return (soreceive(so, 0, uio, 0, 0, 0));
+	CURVNET_SET(so->so_vnet);
+	error = soreceive(so, 0, uio, 0, 0, 0);
+	CURVNET_RESTORE();
+	return (error);
 }
 
 /* ARGSUSED */
@@ -125,6 +129,7 @@ soo_ioctl(struct file *fp, u_long cmd, v
 	struct socket *so = fp->f_data;
 	int error = 0;
 
+	CURVNET_SET(so->so_vnet);
 	switch (cmd) {
 	case FIONBIO:
 		SOCK_LOCK(so);
@@ -205,6 +210,7 @@ soo_ioctl(struct file *fp, u_long cmd, v
 			    (so, cmd, data, 0, td));
 		break;
 	}
+	CURVNET_RESTORE();
 	return (error);
 }
 

Modified: head/sys/kern/uipc_socket.c
==============================================================================
--- head/sys/kern/uipc_socket.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/uipc_socket.c	Tue May  5 10:56:12 2009	(r191816)
@@ -264,7 +264,7 @@ SYSINIT(param, SI_SUB_TUNABLES, SI_ORDER
  * soalloc() returns a socket with a ref count of 0.
  */
 static struct socket *
-soalloc(void)
+soalloc(struct vnet *vnet)
 {
 	struct socket *so;
 
@@ -286,7 +286,8 @@ soalloc(void)
 	so->so_gencnt = ++so_gencnt;
 	++numopensockets;
 #ifdef VIMAGE
-	so->so_vnet = curvnet;
+	++vnet->sockcnt;	/* locked with so_global_mtx */
+	so->so_vnet = vnet;
 #endif
 	mtx_unlock(&so_global_mtx);
 	return (so);
@@ -307,6 +308,9 @@ sodealloc(struct socket *so)
 	mtx_lock(&so_global_mtx);
 	so->so_gencnt = ++so_gencnt;
 	--numopensockets;	/* Could be below, but faster here. */
+#ifdef VIMAGE
+	--so->so_vnet->sockcnt;
+#endif
 	mtx_unlock(&so_global_mtx);
 	if (so->so_rcv.sb_hiwat)
 		(void)chgsbsize(so->so_cred->cr_uidinfo,
@@ -356,7 +360,7 @@ socreate(int dom, struct socket **aso, i
 
 	if (prp->pr_type != type)
 		return (EPROTOTYPE);
-	so = soalloc();
+	so = soalloc(TD_TO_VNET(td));
 	if (so == NULL)
 		return (ENOBUFS);
 
@@ -382,7 +386,9 @@ socreate(int dom, struct socket **aso, i
 	 * Auto-sizing of socket buffers is managed by the protocols and
 	 * the appropriate flags must be set in the pru_attach function.
 	 */
+	CURVNET_SET(so->so_vnet);
 	error = (*prp->pr_usrreqs->pru_attach)(so, proto, td);
+	CURVNET_RESTORE();
 	if (error) {
 		KASSERT(so->so_count == 1, ("socreate: so_count %d",
 		    so->so_count));
@@ -424,7 +430,8 @@ sonewconn(struct socket *head, int conns
 	if (over)
 #endif
 		return (NULL);
-	so = soalloc();
+	VNET_ASSERT(head->so_vnet);
+	so = soalloc(head->so_vnet);
 	if (so == NULL)
 		return (NULL);
 	if ((head->so_options & SO_ACCEPTFILTER) != 0)
@@ -496,8 +503,12 @@ sonewconn(struct socket *head, int conns
 int
 sobind(struct socket *so, struct sockaddr *nam, struct thread *td)
 {
+	int error;
 
-	return ((*so->so_proto->pr_usrreqs->pru_bind)(so, nam, td));
+	CURVNET_SET(so->so_vnet);
+	error = (*so->so_proto->pr_usrreqs->pru_bind)(so, nam, td);
+	CURVNET_RESTORE();
+	return error;
 }
 
 /*
@@ -645,6 +656,7 @@ soclose(struct socket *so)
 
 	KASSERT(!(so->so_state & SS_NOFDREF), ("soclose: SS_NOFDREF on enter"));
 
+	CURVNET_SET(so->so_vnet);
 	funsetown(&so->so_sigio);
 	if (so->so_state & SS_ISCONNECTED) {
 		if ((so->so_state & SS_ISDISCONNECTING) == 0) {
@@ -696,6 +708,7 @@ drop:
 	KASSERT((so->so_state & SS_NOFDREF) == 0, ("soclose: NOFDREF"));
 	so->so_state |= SS_NOFDREF;
 	sorele(so);
+	CURVNET_RESTORE();
 	return (error);
 }
 
@@ -771,7 +784,9 @@ soconnect(struct socket *so, struct sock
 		 * biting us.
 		 */
 		so->so_error = 0;
+		CURVNET_SET(so->so_vnet);
 		error = (*so->so_proto->pr_usrreqs->pru_connect)(so, nam, td);
+		CURVNET_RESTORE();
 	}
 
 	return (error);
@@ -1287,9 +1302,13 @@ int
 sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
     struct mbuf *top, struct mbuf *control, int flags, struct thread *td)
 {
+	int error;
 
-	return (so->so_proto->pr_usrreqs->pru_sosend(so, addr, uio, top,
-	    control, flags, td));
+	CURVNET_SET(so->so_vnet);
+	error = so->so_proto->pr_usrreqs->pru_sosend(so, addr, uio, top,
+		control, flags, td);
+	CURVNET_RESTORE();
+	return (error);
 }
 
 /*
@@ -2037,6 +2056,7 @@ int
 soshutdown(struct socket *so, int how)
 {
 	struct protosw *pr = so->so_proto;
+	int error;
 
 	if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR))
 		return (EINVAL);
@@ -2045,8 +2065,12 @@ soshutdown(struct socket *so, int how)
 	}
 	if (how != SHUT_WR)
 		sorflush(so);
-	if (how != SHUT_RD)
-		return ((*pr->pr_usrreqs->pru_shutdown)(so));
+	if (how != SHUT_RD) {
+		CURVNET_SET(so->so_vnet);
+		error = (*pr->pr_usrreqs->pru_shutdown)(so);
+		CURVNET_RESTORE();
+		return (error);
+	}
 	return (0);
 }
 
@@ -2070,6 +2094,7 @@ sorflush(struct socket *so)
 	 * socket buffer.  Don't let our acquire be interrupted by a signal
 	 * despite any existing socket disposition on interruptable waiting.
 	 */
+	CURVNET_SET(so->so_vnet);
 	socantrcvmore(so);
 	(void) sblock(sb, SBL_WAIT | SBL_NOINTR);
 
@@ -2093,6 +2118,7 @@ sorflush(struct socket *so)
 	if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose != NULL)
 		(*pr->pr_domain->dom_dispose)(asb.sb_mb);
 	sbrelease_internal(&asb, so);
+	CURVNET_RESTORE();
 }
 
 /*

Modified: head/sys/kern/uipc_syscalls.c
==============================================================================
--- head/sys/kern/uipc_syscalls.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/uipc_syscalls.c	Tue May  5 10:56:12 2009	(r191816)
@@ -64,6 +64,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/sysctl.h>
 #include <sys/uio.h>
 #include <sys/vnode.h>
+#include <sys/vimage.h>
 #ifdef KTRACE
 #include <sys/ktrace.h>
 #endif
@@ -264,7 +265,9 @@ listen(td, uap)
 		if (error)
 			goto done;
 #endif
+		CURVNET_SET(so->so_vnet);
 		error = solisten(so, uap->backlog, td);
+		CURVNET_RESTORE();
 #ifdef MAC
 done:
 #endif
@@ -429,7 +432,9 @@ kern_accept(struct thread *td, int s, st
 	tmp = fflag & FASYNC;
 	(void) fo_ioctl(nfp, FIOASYNC, &tmp, td->td_ucred, td);
 	sa = 0;
+	CURVNET_SET(so->so_vnet);
 	error = soaccept(so, &sa);
+	CURVNET_RESTORE();
 	if (error) {
 		/*
 		 * return a namelen of zero for older code which might
@@ -976,9 +981,11 @@ kern_recvit(td, s, mp, fromseg, controlp
 		ktruio = cloneuio(&auio);
 #endif
 	len = auio.uio_resid;
+	CURVNET_SET(so->so_vnet);
 	error = soreceive(so, &fromsa, &auio, (struct mbuf **)0,
 	    (mp->msg_control || controlp) ? &control : (struct mbuf **)0,
 	    &mp->msg_flags);
+	CURVNET_RESTORE();
 	if (error) {
 		if (auio.uio_resid != (int)len && (error == ERESTART ||
 		    error == EINTR || error == EWOULDBLOCK))
@@ -1322,7 +1329,9 @@ kern_setsockopt(td, s, level, name, val,
 	error = getsock(td->td_proc->p_fd, s, &fp, NULL);
 	if (error == 0) {
 		so = fp->f_data;
+		CURVNET_SET(so->so_vnet);
 		error = sosetopt(so, &sopt);
+		CURVNET_RESTORE();
 		fdrop(fp, td);
 	}
 	return(error);
@@ -1400,7 +1409,9 @@ kern_getsockopt(td, s, level, name, val,
 	error = getsock(td->td_proc->p_fd, s, &fp, NULL);
 	if (error == 0) {
 		so = fp->f_data;
+		CURVNET_SET(so->so_vnet);
 		error = sogetopt(so, &sopt);
+		CURVNET_RESTORE();
 		*valsize = sopt.sopt_valsize;
 		fdrop(fp, td);
 	}
@@ -1463,7 +1474,9 @@ kern_getsockname(struct thread *td, int 
 		return (error);
 	so = fp->f_data;
 	*sa = NULL;
+	CURVNET_SET(so->so_vnet);
 	error = (*so->so_proto->pr_usrreqs->pru_sockaddr)(so, sa);
+	CURVNET_RESTORE();
 	if (error)
 		goto bad;
 	if (*sa == NULL)
@@ -1564,7 +1577,9 @@ kern_getpeername(struct thread *td, int 
 		goto done;
 	}
 	*sa = NULL;
+	CURVNET_SET(so->so_vnet);
 	error = (*so->so_proto->pr_usrreqs->pru_peeraddr)(so, sa);
+	CURVNET_RESTORE();
 	if (error)
 		goto bad;
 	if (*sa == NULL)
@@ -2176,9 +2191,11 @@ retry_space:
 				goto done;
 			}
 			SOCKBUF_UNLOCK(&so->so_snd);
+			CURVNET_SET(so->so_vnet);
 			/* Avoid error aliasing. */
 			err = (*so->so_proto->pr_usrreqs->pru_send)
 				    (so, 0, m, NULL, NULL, td);
+			CURVNET_RESTORE();
 			if (err == 0) {
 				/*
 				 * We need two counters to get the

Modified: head/sys/kern/uipc_usrreq.c
==============================================================================
--- head/sys/kern/uipc_usrreq.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/kern/uipc_usrreq.c	Tue May  5 10:56:12 2009	(r191816)
@@ -90,6 +90,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/un.h>
 #include <sys/unpcb.h>
 #include <sys/vnode.h>
+#include <sys/vimage.h>
 
 #ifdef DDB
 #include <ddb/ddb.h>
@@ -1647,6 +1648,10 @@ static void
 unp_init(void)
 {
 
+#ifdef VIMAGE
+	if (!IS_DEFAULT_VNET(curvnet))
+		return;
+#endif
 	unp_zone = uma_zcreate("unpcb", sizeof(struct unpcb), NULL, NULL,
 	    NULL, NULL, UMA_ALIGN_PTR, 0);
 	if (unp_zone == NULL)

Modified: head/sys/net/bpf.c
==============================================================================
--- head/sys/net/bpf.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/bpf.c	Tue May  5 10:56:12 2009	(r191816)
@@ -873,11 +873,10 @@ bpfwrite(struct cdev *dev, struct uio *u
 	m->m_len -= hlen;
 	m->m_data += hlen;	/* XXX */
 
+	CURVNET_SET(ifp->if_vnet);
 #ifdef MAC
 	BPFD_LOCK(d);
-	CURVNET_SET(ifp->if_vnet);
 	mac_bpfdesc_create_mbuf(d, m);
-	CURVNET_RESTORE();
 	if (mc != NULL)
 		mac_bpfdesc_create_mbuf(d, mc);
 	BPFD_UNLOCK(d);
@@ -893,6 +892,7 @@ bpfwrite(struct cdev *dev, struct uio *u
 		else
 			m_freem(mc);
 	}
+	CURVNET_RESTORE();
 
 	return (error);
 }

Modified: head/sys/net/if.c
==============================================================================
--- head/sys/net/if.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/if.c	Tue May  5 10:56:12 2009	(r191816)
@@ -53,6 +53,7 @@
 #include <sys/kernel.h>
 #include <sys/lock.h>
 #include <sys/refcount.h>
+#include <sys/module.h>
 #include <sys/rwlock.h>
 #include <sys/sockio.h>
 #include <sys/syslog.h>
@@ -126,7 +127,6 @@ static void	if_attachdomain(void *);
 static void	if_attachdomain1(struct ifnet *);
 static int	ifconf(u_long, caddr_t);
 static void	if_freemulti(struct ifmultiaddr *);
-static void	if_grow(void);
 static void	if_init(void *);
 static void	if_check(void *);
 static void	if_route(struct ifnet *, int flag, int fam);
@@ -202,7 +202,7 @@ MALLOC_DEFINE(M_IFNET, "ifnet", "interfa
 MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
 MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");
 
-static struct ifnet *
+struct ifnet *
 ifnet_byindex_locked(u_short idx)
 {
 	INIT_VNET_NET(curvnet);
@@ -239,7 +239,7 @@ ifnet_byindex_ref(u_short idx)
 	return (ifp);
 }
 
-static void
+void
 ifnet_setbyindex(u_short idx, struct ifnet *ifp)
 {
 	INIT_VNET_NET(curvnet);
@@ -445,7 +445,7 @@ vnet_net_iattach(const void *unused __un
 	return (0);
 }
 
-static void
+void
 if_grow(void)
 {
 	INIT_VNET_NET(curvnet);
@@ -696,11 +696,13 @@ if_attach(struct ifnet *ifp)
 	mac_ifnet_create(ifp);
 #endif
 
-	ifdev_setbyindex(ifp->if_index, make_dev(&net_cdevsw,
-	    ifp->if_index, UID_ROOT, GID_WHEEL, 0600, "%s/%s",
-	    net_cdevsw.d_name, ifp->if_xname));
-	make_dev_alias(ifdev_byindex(ifp->if_index), "%s%d",
-	    net_cdevsw.d_name, ifp->if_index);
+	if (IS_DEFAULT_VNET(curvnet)) {
+		ifdev_setbyindex(ifp->if_index, make_dev(&net_cdevsw,
+		    ifp->if_index, UID_ROOT, GID_WHEEL, 0600, "%s/%s",
+		    net_cdevsw.d_name, ifp->if_xname));
+		make_dev_alias(ifdev_byindex(ifp->if_index), "%s%d",
+		    net_cdevsw.d_name, ifp->if_index);
+	}
 
 	ifq_attach(&ifp->if_snd, ifp);
 
@@ -742,13 +744,17 @@ if_attach(struct ifnet *ifp)
 
 	IFNET_WLOCK();
 	TAILQ_INSERT_TAIL(&V_ifnet, ifp, if_link);
+#ifdef VIMAGE
+	curvnet->ifccnt++;
+#endif
 	IFNET_WUNLOCK();
 
 	if (domain_init_status >= 2)
 		if_attachdomain1(ifp);
 
 	EVENTHANDLER_INVOKE(ifnet_arrival_event, ifp);
-	devctl_notify("IFNET", ifp->if_xname, "ATTACH", NULL);
+	if (IS_DEFAULT_VNET(curvnet))
+		devctl_notify("IFNET", ifp->if_xname, "ATTACH", NULL);
 
 	/* Announce the interface. */
 	rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
@@ -895,6 +901,10 @@ if_detach(struct ifnet *ifp)
 			found = 1;
 			break;
 		}
+#ifdef VIMAGE
+	if (found)
+		curvnet->ifccnt--;
+#endif
 	IFNET_WUNLOCK();
 	if (!found)
 		return;
@@ -943,7 +953,8 @@ if_detach(struct ifnet *ifp)
 	 * Clean up all addresses.
 	 */
 	ifp->if_addr = NULL;
-	destroy_dev(ifdev_byindex(ifp->if_index));
+	if (IS_DEFAULT_VNET(curvnet))
+		destroy_dev(ifdev_byindex(ifp->if_index));
 	ifdev_setbyindex(ifp->if_index, NULL);	
 
 	/* We can now free link ifaddr. */
@@ -972,7 +983,8 @@ if_detach(struct ifnet *ifp)
 	/* Announce that the interface is gone. */
 	rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
 	EVENTHANDLER_INVOKE(ifnet_departure_event, ifp);
-	devctl_notify("IFNET", ifp->if_xname, "DETACH", NULL);
+	if (IS_DEFAULT_VNET(curvnet))
+		devctl_notify("IFNET", ifp->if_xname, "DETACH", NULL);
 	if_delgroups(ifp);
 
 	IF_AFDATA_LOCK(ifp);
@@ -1701,8 +1713,10 @@ do_link_state_change(void *arg, int pend
 		(*lagg_linkstate_p)(ifp, link_state);
 	}
 
-	devctl_notify("IFNET", ifp->if_xname,
-	    (link_state == LINK_STATE_UP) ? "LINK_UP" : "LINK_DOWN", NULL);
+	if (IS_DEFAULT_VNET(curvnet))
+		devctl_notify("IFNET", ifp->if_xname,
+		    (link_state == LINK_STATE_UP) ? "LINK_UP" : "LINK_DOWN",
+		    NULL);
 	if (pending > 1)
 		if_printf(ifp, "%d link states coalesced\n", pending);
 	if (log_link_state_change)

Modified: head/sys/net/if_clone.c
==============================================================================
--- head/sys/net/if_clone.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/if_clone.c	Tue May  5 10:56:12 2009	(r191816)
@@ -39,6 +39,7 @@
 #include <sys/systm.h>
 #include <sys/types.h>
 #include <sys/socket.h>
+#include <sys/vimage.h>
 
 #include <net/if.h>
 #include <net/if_clone.h>
@@ -49,6 +50,7 @@
 #include <net/if_var.h>
 #include <net/radix.h>
 #include <net/route.h>
+#include <net/vnet.h>
 
 static void	if_clone_free(struct if_clone *ifc);
 static int	if_clone_createif(struct if_clone *ifc, char *name, size_t len,
@@ -203,15 +205,14 @@ if_clone_destroyif(struct if_clone *ifc,
 {
 	int err;
 
-	if (ifc->ifc_destroy == NULL) {
-		err = EOPNOTSUPP;
-		goto done;
-	}
+	if (ifc->ifc_destroy == NULL)
+		return(EOPNOTSUPP);
 
 	IF_CLONE_LOCK(ifc);
 	IFC_IFLIST_REMOVE(ifc, ifp);
 	IF_CLONE_UNLOCK(ifc);
 
+	CURVNET_SET_QUIET(ifp->if_vnet);
 	if_delgroup(ifp, ifc->ifc_name);
 
 	err =  (*ifc->ifc_destroy)(ifc, ifp);
@@ -223,8 +224,7 @@ if_clone_destroyif(struct if_clone *ifc,
 		IFC_IFLIST_INSERT(ifc, ifp);
 		IF_CLONE_UNLOCK(ifc);
 	}
-
-done:
+	CURVNET_RESTORE();
 	return (err);
 }
 

Modified: head/sys/net/if_ethersubr.c
==============================================================================
--- head/sys/net/if_ethersubr.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/if_ethersubr.c	Tue May  5 10:56:12 2009	(r191816)
@@ -602,6 +602,8 @@ ether_input(struct ifnet *ifp, struct mb
 	}
 #endif
 
+	CURVNET_SET_QUIET(ifp->if_vnet);
+
 	if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
 		if (ETHER_IS_BROADCAST(eh->ether_dhost))
 			m->m_flags |= M_BCAST;
@@ -638,6 +640,7 @@ ether_input(struct ifnet *ifp, struct mb
 	/* Allow monitor mode to claim this frame, after stats are updated. */
 	if (ifp->if_flags & IFF_MONITOR) {
 		m_freem(m);
+		CURVNET_RESTORE();
 		return;
 	}
 
@@ -686,8 +689,10 @@ ether_input(struct ifnet *ifp, struct mb
 		    ("%s: ng_ether_input_p is NULL", __func__));
 		m->m_flags &= ~M_PROMISC;
 		(*ng_ether_input_p)(ifp, &m);
-		if (m == NULL)
+		if (m == NULL) {
+			CURVNET_RESTORE();
 			return;
+		}
 	}
 
 	/*
@@ -698,8 +703,10 @@ ether_input(struct ifnet *ifp, struct mb
 	if (ifp->if_bridge != NULL) {
 		m->m_flags &= ~M_PROMISC;
 		BRIDGE_INPUT(ifp, m);
-		if (m == NULL)
+		if (m == NULL) {
+			CURVNET_RESTORE();
 			return;
+		}
 	}
 
 #ifdef DEV_CARP
@@ -735,6 +742,7 @@ ether_input(struct ifnet *ifp, struct mb
 		random_harvest(m, 16, 3, 0, RANDOM_NET);
 
 	ether_demux(ifp, m);
+	CURVNET_RESTORE();
 }
 
 /*

Modified: head/sys/net/if_var.h
==============================================================================
--- head/sys/net/if_var.h	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/if_var.h	Tue May  5 10:56:12 2009	(r191816)
@@ -731,7 +731,9 @@ struct ifindex_entry {
  * to call ifnet_byindex() instead if ifnet_byindex_ref().
  */
 struct ifnet	*ifnet_byindex(u_short idx);
+struct ifnet	*ifnet_byindex_locked(u_short idx);
 struct ifnet	*ifnet_byindex_ref(u_short idx);
+void ifnet_setbyindex(u_short idx, struct ifnet *ifp);
 
 /*
  * Given the index, ifaddr_byindex() returns the one and only
@@ -755,6 +757,7 @@ int	if_allmulti(struct ifnet *, int);
 struct	ifnet* if_alloc(u_char);
 void	if_attach(struct ifnet *);
 void	if_dead(struct ifnet *);
+void	if_grow(void);
 int	if_delmulti(struct ifnet *, struct sockaddr *);
 void	if_delmulti_ifma(struct ifmultiaddr *);
 void	if_detach(struct ifnet *);

Modified: head/sys/net/netisr.c
==============================================================================
--- head/sys/net/netisr.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/netisr.c	Tue May  5 10:56:12 2009	(r191816)
@@ -43,6 +43,7 @@
 #include <sys/resourcevar.h>
 #include <sys/sysctl.h>
 #include <sys/unistd.h>
+#include <sys/vimage.h>
 #include <machine/atomic.h>
 #include <machine/cpu.h>
 #include <machine/stdarg.h>
@@ -142,7 +143,10 @@ netisr_processqueue(struct netisr *ni)
 		IF_DEQUEUE(ni->ni_queue, m);
 		if (m == NULL)
 			break;
+		VNET_ASSERT(m->m_pkthdr.rcvif != NULL);
+		CURVNET_SET(m->m_pkthdr.rcvif->if_vnet);
 		ni->ni_handler(m);
+		CURVNET_RESTORE();
 	}
 }
 

Modified: head/sys/net/rtsock.c
==============================================================================
--- head/sys/net/rtsock.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net/rtsock.c	Tue May  5 10:56:12 2009	(r191816)
@@ -1206,6 +1206,7 @@ rt_ifannouncemsg(struct ifnet *ifp, int 
 static void
 rt_dispatch(struct mbuf *m, const struct sockaddr *sa)
 {
+	INIT_VNET_NET(curvnet);
 	struct m_tag *tag;
 
 	/*
@@ -1223,6 +1224,14 @@ rt_dispatch(struct mbuf *m, const struct
 		*(unsigned short *)(tag + 1) = sa->sa_family;
 		m_tag_prepend(m, tag);
 	}
+#ifdef VIMAGE
+	if (V_loif)
+		m->m_pkthdr.rcvif = V_loif;
+	else {
+		m_freem(m);
+		return;
+	}
+#endif
 	netisr_queue(NETISR_ROUTE, m);	/* mbuf is free'd on failure. */
 }
 

Modified: head/sys/net80211/ieee80211_freebsd.c
==============================================================================
--- head/sys/net80211/ieee80211_freebsd.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/net80211/ieee80211_freebsd.c	Tue May  5 10:56:12 2009	(r191816)
@@ -41,6 +41,7 @@ __FBSDID("$FreeBSD$");
 #include <sys/sysctl.h>
 
 #include <sys/socket.h>
+#include <sys/vimage.h>
 
 #include <net/if.h>
 #include <net/if_dl.h>
@@ -498,9 +499,11 @@ notify_macaddr(struct ifnet *ifp, int op
 {
 	struct ieee80211_join_event iev;
 
+	CURVNET_SET(ifp->if_vnet);
 	memset(&iev, 0, sizeof(iev));
 	IEEE80211_ADDR_COPY(iev.iev_addr, mac);
 	rt_ieee80211msg(ifp, op, &iev, sizeof(iev));
+	CURVNET_RESTORE();
 }
 
 void
@@ -509,6 +512,7 @@ ieee80211_notify_node_join(struct ieee80
 	struct ieee80211vap *vap = ni->ni_vap;
 	struct ifnet *ifp = vap->iv_ifp;
 
+	CURVNET_SET_QUIET(ifp->if_vnet);
 	IEEE80211_NOTE(vap, IEEE80211_MSG_NODE, ni, "%snode join",
 	    (ni == vap->iv_bss) ? "bss " : "");
 
@@ -520,6 +524,7 @@ ieee80211_notify_node_join(struct ieee80
 		notify_macaddr(ifp, newassoc ?
 		    RTM_IEEE80211_JOIN : RTM_IEEE80211_REJOIN, ni->ni_macaddr);
 	}
+	CURVNET_RESTORE();
 }
 
 void
@@ -528,6 +533,7 @@ ieee80211_notify_node_leave(struct ieee8
 	struct ieee80211vap *vap = ni->ni_vap;
 	struct ifnet *ifp = vap->iv_ifp;
 
+	CURVNET_SET_QUIET(ifp->if_vnet);
 	IEEE80211_NOTE(vap, IEEE80211_MSG_NODE, ni, "%snode leave",
 	    (ni == vap->iv_bss) ? "bss " : "");
 
@@ -538,6 +544,7 @@ ieee80211_notify_node_leave(struct ieee8
 		/* fire off wireless event station leaving */
 		notify_macaddr(ifp, RTM_IEEE80211_LEAVE, ni->ni_macaddr);
 	}
+	CURVNET_RESTORE();
 }
 
 void
@@ -548,7 +555,9 @@ ieee80211_notify_scan_done(struct ieee80
 	IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s\n", "notify scan done");
 
 	/* dispatch wireless event indicating scan completed */
+	CURVNET_SET(ifp->if_vnet);
 	rt_ieee80211msg(ifp, RTM_IEEE80211_SCAN, NULL, 0);
+	CURVNET_RESTORE();
 }
 
 void
@@ -576,7 +585,9 @@ ieee80211_notify_replay_failure(struct i
 			iev.iev_keyix = k->wk_keyix;
 		iev.iev_keyrsc = k->wk_keyrsc[0];	/* XXX need tid */
 		iev.iev_rsc = rsc;
+		CURVNET_SET(ifp->if_vnet);
 		rt_ieee80211msg(ifp, RTM_IEEE80211_REPLAY, &iev, sizeof(iev));
+		CURVNET_RESTORE();
 	}
 }
 
@@ -597,7 +608,9 @@ ieee80211_notify_michael_failure(struct 
 		IEEE80211_ADDR_COPY(iev.iev_src, wh->i_addr2);
 		iev.iev_cipher = IEEE80211_CIPHER_TKIP;
 		iev.iev_keyix = keyix;
+		CURVNET_SET(ifp->if_vnet);
 		rt_ieee80211msg(ifp, RTM_IEEE80211_MICHAEL, &iev, sizeof(iev));
+		CURVNET_RESTORE();
 	}
 }
 

Modified: head/sys/netgraph/netgraph.h
==============================================================================
--- head/sys/netgraph/netgraph.h	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/netgraph/netgraph.h	Tue May  5 10:56:12 2009	(r191816)
@@ -352,6 +352,7 @@ struct ng_node {
 	LIST_ENTRY(ng_node)	  nd_idnodes;	/* ID hash collision list */
 	struct	ng_queue	  nd_input_queue; /* input queue for locking */
 	int	nd_refs;		/* # of references to this node */
+	struct	vnet		 *nd_vnet;	/* network stack instance */
 #ifdef	NETGRAPH_DEBUG /*----------------------------------------------*/
 #define ND_MAGIC 0x59264837
 	int	nd_magic;

Modified: head/sys/netinet/if_ether.c
==============================================================================
--- head/sys/netinet/if_ether.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/netinet/if_ether.c	Tue May  5 10:56:12 2009	(r191816)
@@ -143,10 +143,12 @@ arp_ifscrub(struct ifnet *ifp, uint32_t 
 	addr4.sin_len    = sizeof(addr4);
 	addr4.sin_family = AF_INET;
 	addr4.sin_addr.s_addr = addr;
+	CURVNET_SET(ifp->if_vnet);
 	IF_AFDATA_LOCK(ifp);
 	lla_lookup(LLTABLE(ifp), (LLE_DELETE | LLE_IFADDR),
 	    (struct sockaddr *)&addr4);
 	IF_AFDATA_UNLOCK(ifp);
+	CURVNET_RESTORE();
 }
 #endif
 

Modified: head/sys/netinet/igmp.c
==============================================================================
--- head/sys/netinet/igmp.c	Tue May  5 10:46:49 2009	(r191815)
+++ head/sys/netinet/igmp.c	Tue May  5 10:56:12 2009	(r191816)
@@ -1117,6 +1117,9 @@ igmp_input_v3_group_query(struct in_mult
 
 	nsrc = ntohs(igmpv3->igmp_numsrc);
 
+	if (!IS_DEFAULT_VNET(curvnet))
+		return (retval);
+
 	/*
 	 * Deal with group-specific queries upfront.
 	 * If any group query is already pending, purge any recorded
@@ -3372,7 +3375,7 @@ igmp_intr(struct mbuf *m)
 	 * indexes to guard against interface detach, they are
 	 * unique to each VIMAGE and must be retrieved.
 	 */
-	CURVNET_SET(m->m_pkthdr.header);
+	CURVNET_SET((struct vnet *)(m->m_pkthdr.header));
 	INIT_VNET_NET(curvnet);
 	INIT_VNET_INET(curvnet);
 	ifindex = igmp_restore_context(m);
@@ -3654,9 +3657,7 @@ igmp_modevent(module_t mod, int type, vo
 	break;
     case MOD_UNLOAD:
 #ifndef VIMAGE_GLOBALS

*** DIFF OUTPUT TRUNCATED AT 1000 LINES ***


More information about the svn-src-all mailing list