2005-01-07 01:45:51 +00:00
|
|
|
/*-
|
2017-11-20 19:43:44 +00:00
|
|
|
* SPDX-License-Identifier: BSD-3-Clause
|
|
|
|
*
|
1994-05-24 10:09:53 +00:00
|
|
|
* Copyright (c) 1982, 1986, 1988, 1993
|
2008-07-05 18:03:39 +00:00
|
|
|
* The Regents of the University of California.
|
|
|
|
* All rights reserved.
|
1994-05-24 10:09:53 +00:00
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
2017-02-28 23:42:47 +00:00
|
|
|
* 3. Neither the name of the University nor the names of its contributors
|
1994-05-24 10:09:53 +00:00
|
|
|
* may be used to endorse or promote products derived from this software
|
|
|
|
* without specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
|
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
|
|
* SUCH DAMAGE.
|
|
|
|
*
|
1995-09-21 19:59:43 +00:00
|
|
|
* @(#)raw_ip.c 8.7 (Berkeley) 5/15/95
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
|
|
|
|
2007-10-07 20:44:24 +00:00
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
2011-04-20 08:03:22 +00:00
|
|
|
#include "opt_inet.h"
|
1999-12-22 19:13:38 +00:00
|
|
|
#include "opt_inet6.h"
|
|
|
|
#include "opt_ipsec.h"
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/param.h>
|
2004-04-26 19:46:52 +00:00
|
|
|
#include <sys/jail.h>
|
1997-02-18 20:46:36 +00:00
|
|
|
#include <sys/kernel.h>
|
2016-01-09 09:34:39 +00:00
|
|
|
#include <sys/eventhandler.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/lock.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/malloc.h>
|
|
|
|
#include <sys/mbuf.h>
|
2006-11-06 13:42:10 +00:00
|
|
|
#include <sys/priv.h>
|
2001-10-05 07:06:32 +00:00
|
|
|
#include <sys/proc.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/protosw.h>
|
2015-07-29 08:12:05 +00:00
|
|
|
#include <sys/rmlock.h>
|
Conditionally compile out V_ globals while instantiating the appropriate
container structures, depending on VIMAGE_GLOBALS compile time option.
Make VIMAGE_GLOBALS a new compile-time option, which by default will not
be defined, resulting in instatiations of global variables selected for
V_irtualization (enclosed in #ifdef VIMAGE_GLOBALS blocks) to be
effectively compiled out. Instantiate new global container structures
to hold V_irtualized variables: vnet_net_0, vnet_inet_0, vnet_inet6_0,
vnet_ipsec_0, vnet_netgraph_0, and vnet_gif_0.
Update the VSYM() macro so that depending on VIMAGE_GLOBALS the V_
macros resolve either to the original globals, or to fields inside
container structures, i.e. effectively
#ifdef VIMAGE_GLOBALS
#define V_rt_tables rt_tables
#else
#define V_rt_tables vnet_net_0._rt_tables
#endif
Update SYSCTL_V_*() macros to operate either on globals or on fields
inside container structs.
Extend the internal kldsym() lookups with the ability to resolve
selected fields inside the virtualization container structs. This
applies only to the fields which are explicitly registered for kldsym()
visibility via VNET_MOD_DECLARE() and vnet_mod_register(), currently
this is done only in sys/net/if.c.
Fix a few broken instances of MODULE_GLOBAL() macro use in SCTP code,
and modify the MODULE_GLOBAL() macro to resolve to V_ macros, which in
turn result in proper code being generated depending on VIMAGE_GLOBALS.
De-virtualize local static variables in sys/contrib/pf/net/pf_subr.c
which were prematurely V_irtualized by automated V_ prepending scripts
during earlier merging steps. PF virtualization will be done
separately, most probably after next PF import.
Convert a few variable initializations at instantiation to
initialization in init functions, most notably in ipfw. Also convert
TUNABLE_INT() initializers for V_ variables to TUNABLE_FETCH_INT() in
initializer functions.
Discussed at: devsummit Strassburg
Reviewed by: bz, julian
Approved by: julian (mentor)
Obtained from: //depot/projects/vimage-commit2/...
X-MFC after: never
Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-12-10 23:12:39 +00:00
|
|
|
#include <sys/rwlock.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/signalvar.h>
|
1997-02-18 20:46:36 +00:00
|
|
|
#include <sys/socket.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <sys/socketvar.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/sx.h>
|
1997-02-18 20:46:36 +00:00
|
|
|
#include <sys/sysctl.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <sys/systm.h>
|
1998-03-28 10:18:26 +00:00
|
|
|
|
2002-03-20 05:48:55 +00:00
|
|
|
#include <vm/uma.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
#include <net/if.h>
|
2013-10-26 17:58:36 +00:00
|
|
|
#include <net/if_var.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <net/route.h>
|
2008-12-02 21:37:28 +00:00
|
|
|
#include <net/vnet.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
|
|
|
#include <netinet/in.h>
|
|
|
|
#include <netinet/in_systm.h>
|
1995-10-21 02:12:20 +00:00
|
|
|
#include <netinet/in_pcb.h>
|
|
|
|
#include <netinet/in_var.h>
|
2011-05-20 19:12:20 +00:00
|
|
|
#include <netinet/if_ether.h>
|
2002-04-30 01:54:54 +00:00
|
|
|
#include <netinet/ip.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
#include <netinet/ip_var.h>
|
|
|
|
#include <netinet/ip_mroute.h>
|
2016-05-25 15:54:21 +00:00
|
|
|
#include <netinet/ip_icmp.h>
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2017-02-06 08:49:57 +00:00
|
|
|
#include <netipsec/ipsec_support.h>
|
2002-10-16 02:25:05 +00:00
|
|
|
|
2014-08-15 02:43:02 +00:00
|
|
|
#include <machine/stdarg.h>
|
2006-10-22 11:52:19 +00:00
|
|
|
#include <security/mac/mac_framework.h>
|
|
|
|
|
2011-04-27 19:32:27 +00:00
|
|
|
VNET_DEFINE(int, ip_defttl) = IPDEFTTL;
|
2014-11-07 09:39:05 +00:00
|
|
|
SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_VNET | CTLFLAG_RW,
|
2011-04-27 19:32:27 +00:00
|
|
|
&VNET_NAME(ip_defttl), 0,
|
|
|
|
"Maximum TTL on IP packets");
|
|
|
|
|
Build on Jeff Roberson's linker-set based dynamic per-CPU allocator
(DPCPU), as suggested by Peter Wemm, and implement a new per-virtual
network stack memory allocator. Modify vnet to use the allocator
instead of monolithic global container structures (vinet, ...). This
change solves many binary compatibility problems associated with
VIMAGE, and restores ELF symbols for virtualized global variables.
Each virtualized global variable exists as a "reference copy", and also
once per virtual network stack. Virtualized global variables are
tagged at compile-time, placing the in a special linker set, which is
loaded into a contiguous region of kernel memory. Virtualized global
variables in the base kernel are linked as normal, but those in modules
are copied and relocated to a reserved portion of the kernel's vnet
region with the help of a the kernel linker.
Virtualized global variables exist in per-vnet memory set up when the
network stack instance is created, and are initialized statically from
the reference copy. Run-time access occurs via an accessor macro, which
converts from the current vnet and requested symbol to a per-vnet
address. When "options VIMAGE" is not compiled into the kernel, normal
global ELF symbols will be used instead and indirection is avoided.
This change restores static initialization for network stack global
variables, restores support for non-global symbols and types, eliminates
the need for many subsystem constructors, eliminates large per-subsystem
structures that caused many binary compatibility issues both for
monitoring applications (netstat) and kernel modules, removes the
per-function INIT_VNET_*() macros throughout the stack, eliminates the
need for vnet_symmap ksym(2) munging, and eliminates duplicate
definitions of virtualized globals under VIMAGE_GLOBALS.
Bump __FreeBSD_version and update UPDATING.
Portions submitted by: bz
Reviewed by: bz, zec
Discussed with: gnn, jamie, jeff, jhb, julian, sam
Suggested by: peter
Approved by: re (kensmith)
2009-07-14 22:48:30 +00:00
|
|
|
VNET_DEFINE(struct inpcbhead, ripcb);
|
|
|
|
VNET_DEFINE(struct inpcbinfo, ripcbinfo);
|
|
|
|
|
2009-07-16 21:13:04 +00:00
|
|
|
#define V_ripcb VNET(ripcb)
|
|
|
|
#define V_ripcbinfo VNET(ripcbinfo)
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2009-06-05 13:44:30 +00:00
|
|
|
/*
|
2010-01-07 10:39:15 +00:00
|
|
|
* Control and data hooks for ipfw, dummynet, divert and so on.
|
2009-06-05 13:44:30 +00:00
|
|
|
* The data hooks are not used here but it is convenient
|
|
|
|
* to keep them all in one place.
|
|
|
|
*/
|
2009-10-11 05:59:43 +00:00
|
|
|
VNET_DEFINE(ip_fw_chk_ptr_t, ip_fw_chk_ptr) = NULL;
|
|
|
|
VNET_DEFINE(ip_fw_ctl_ptr_t, ip_fw_ctl_ptr) = NULL;
|
2010-01-07 10:39:15 +00:00
|
|
|
|
|
|
|
int (*ip_dn_ctl_ptr)(struct sockopt *);
|
2019-03-14 22:32:50 +00:00
|
|
|
int (*ip_dn_io_ptr)(struct mbuf **, struct ip_fw_args *);
|
2019-03-14 22:23:09 +00:00
|
|
|
void (*ip_divert_ptr)(struct mbuf *, bool);
|
2019-03-14 22:30:05 +00:00
|
|
|
int (*ng_ipfw_input_p)(struct mbuf **, struct ip_fw_args *, bool);
|
2001-10-05 05:45:27 +00:00
|
|
|
|
2011-04-20 08:03:22 +00:00
|
|
|
#ifdef INET
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* Hooks for multicast routing. They all default to NULL, so leave them not
|
|
|
|
* initialized and rely on BSS being set to 0.
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
*/
|
|
|
|
|
2008-07-05 18:03:39 +00:00
|
|
|
/*
|
|
|
|
* The socket used to communicate with the multicast routing daemon.
|
|
|
|
*/
|
Build on Jeff Roberson's linker-set based dynamic per-CPU allocator
(DPCPU), as suggested by Peter Wemm, and implement a new per-virtual
network stack memory allocator. Modify vnet to use the allocator
instead of monolithic global container structures (vinet, ...). This
change solves many binary compatibility problems associated with
VIMAGE, and restores ELF symbols for virtualized global variables.
Each virtualized global variable exists as a "reference copy", and also
once per virtual network stack. Virtualized global variables are
tagged at compile-time, placing the in a special linker set, which is
loaded into a contiguous region of kernel memory. Virtualized global
variables in the base kernel are linked as normal, but those in modules
are copied and relocated to a reserved portion of the kernel's vnet
region with the help of a the kernel linker.
Virtualized global variables exist in per-vnet memory set up when the
network stack instance is created, and are initialized statically from
the reference copy. Run-time access occurs via an accessor macro, which
converts from the current vnet and requested symbol to a per-vnet
address. When "options VIMAGE" is not compiled into the kernel, normal
global ELF symbols will be used instead and indirection is avoided.
This change restores static initialization for network stack global
variables, restores support for non-global symbols and types, eliminates
the need for many subsystem constructors, eliminates large per-subsystem
structures that caused many binary compatibility issues both for
monitoring applications (netstat) and kernel modules, removes the
per-function INIT_VNET_*() macros throughout the stack, eliminates the
need for vnet_symmap ksym(2) munging, and eliminates duplicate
definitions of virtualized globals under VIMAGE_GLOBALS.
Bump __FreeBSD_version and update UPDATING.
Portions submitted by: bz
Reviewed by: bz, zec
Discussed with: gnn, jamie, jeff, jhb, julian, sam
Suggested by: peter
Approved by: re (kensmith)
2009-07-14 22:48:30 +00:00
|
|
|
VNET_DEFINE(struct socket *, ip_mrouter);
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
|
2008-07-05 18:03:39 +00:00
|
|
|
/*
|
|
|
|
* The various mrouter and rsvp functions.
|
|
|
|
*/
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
int (*ip_mrouter_set)(struct socket *, struct sockopt *);
|
|
|
|
int (*ip_mrouter_get)(struct socket *, struct sockopt *);
|
|
|
|
int (*ip_mrouter_done)(void);
|
|
|
|
int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
|
2004-08-16 18:32:07 +00:00
|
|
|
struct ip_moptions *);
|
2009-06-21 10:29:31 +00:00
|
|
|
int (*mrt_ioctl)(u_long, caddr_t, int);
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
int (*legal_vif_num)(int);
|
|
|
|
u_long (*ip_mcast_src)(int);
|
|
|
|
|
2014-08-08 01:57:15 +00:00
|
|
|
int (*rsvp_input_p)(struct mbuf **, int *, int);
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
|
|
|
|
void (*ip_rsvp_force_done)(struct socket *);
|
2011-04-20 08:03:22 +00:00
|
|
|
#endif /* INET */
|
|
|
|
|
2016-05-25 13:48:26 +00:00
|
|
|
extern struct protosw inetsw[];
|
|
|
|
|
2011-04-20 08:03:22 +00:00
|
|
|
u_long rip_sendspace = 9216;
|
|
|
|
SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
|
|
|
|
&rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
|
|
|
|
|
|
|
|
u_long rip_recvspace = 9216;
|
|
|
|
SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
|
|
|
|
&rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
|
2008-07-26 17:32:15 +00:00
|
|
|
/*
|
|
|
|
* Hash functions
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define INP_PCBHASH_RAW_SIZE 256
|
|
|
|
#define INP_PCBHASH_RAW(proto, laddr, faddr, mask) \
|
|
|
|
(((proto) + (laddr) + (faddr)) % (mask) + 1)
|
|
|
|
|
2011-04-20 08:03:22 +00:00
|
|
|
#ifdef INET
|
2008-07-26 17:32:15 +00:00
|
|
|
static void
|
|
|
|
rip_inshash(struct inpcb *inp)
|
|
|
|
{
|
|
|
|
struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
|
|
|
|
struct inpcbhead *pcbhash;
|
|
|
|
int hash;
|
|
|
|
|
|
|
|
INP_INFO_WLOCK_ASSERT(pcbinfo);
|
|
|
|
INP_WLOCK_ASSERT(inp);
|
|
|
|
|
2008-07-28 06:57:28 +00:00
|
|
|
if (inp->inp_ip_p != 0 &&
|
|
|
|
inp->inp_laddr.s_addr != INADDR_ANY &&
|
|
|
|
inp->inp_faddr.s_addr != INADDR_ANY) {
|
2008-07-26 17:32:15 +00:00
|
|
|
hash = INP_PCBHASH_RAW(inp->inp_ip_p, inp->inp_laddr.s_addr,
|
|
|
|
inp->inp_faddr.s_addr, pcbinfo->ipi_hashmask);
|
2008-07-28 06:57:28 +00:00
|
|
|
} else
|
2008-07-26 17:32:15 +00:00
|
|
|
hash = 0;
|
|
|
|
pcbhash = &pcbinfo->ipi_hashbase[hash];
|
2018-06-12 22:18:20 +00:00
|
|
|
CK_LIST_INSERT_HEAD(pcbhash, inp, inp_hash);
|
2008-07-26 17:32:15 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
rip_delhash(struct inpcb *inp)
|
|
|
|
{
|
2008-07-28 06:57:28 +00:00
|
|
|
|
|
|
|
INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
|
2008-07-26 17:32:15 +00:00
|
|
|
INP_WLOCK_ASSERT(inp);
|
2008-07-28 06:57:28 +00:00
|
|
|
|
2018-06-12 22:18:20 +00:00
|
|
|
CK_LIST_REMOVE(inp, inp_hash);
|
2008-07-26 17:32:15 +00:00
|
|
|
}
|
2011-04-20 08:03:22 +00:00
|
|
|
#endif /* INET */
|
2008-07-26 17:32:15 +00:00
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
|
|
|
* Raw interface to IP protocol.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
2002-11-20 19:00:54 +00:00
|
|
|
* Initialize raw connection block q.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2006-04-21 09:25:40 +00:00
|
|
|
static void
|
|
|
|
rip_zone_change(void *tag)
|
|
|
|
{
|
|
|
|
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets);
|
2006-04-21 09:25:40 +00:00
|
|
|
}
|
|
|
|
|
2006-07-18 22:34:27 +00:00
|
|
|
static int
|
|
|
|
rip_inpcb_init(void *mem, int size, int flags)
|
|
|
|
{
|
2006-12-29 14:58:18 +00:00
|
|
|
struct inpcb *inp = mem;
|
|
|
|
|
2006-07-18 22:34:27 +00:00
|
|
|
INP_LOCK_INIT(inp, "inp", "rawinp");
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
void
|
2007-05-10 15:58:48 +00:00
|
|
|
rip_init(void)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2007-05-10 15:58:48 +00:00
|
|
|
|
2010-03-14 18:59:11 +00:00
|
|
|
in_pcbinfo_init(&V_ripcbinfo, "rip", &V_ripcb, INP_PCBHASH_RAW_SIZE,
|
2017-05-15 21:58:36 +00:00
|
|
|
1, "ripcb", rip_inpcb_init, IPI_HASHFIELDS_NONE);
|
2008-07-05 18:03:39 +00:00
|
|
|
EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change, NULL,
|
|
|
|
EVENTHANDLER_PRI_ANY);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
Introduce an infrastructure for dismantling vnet instances.
Vnet modules and protocol domains may now register destructor
functions to clean up and release per-module state. The destructor
mechanisms can be triggered by invoking "vimage -d", or a future
equivalent command which will be provided via the new jail framework.
While this patch introduces numerous placeholder destructor functions,
many of those are currently incomplete, thus leaking memory or (even
worse) failing to stop all running timers. Many of such issues are
already known and will be incrementaly fixed over the next weeks in
smaller incremental commits.
Apart from introducing new fields in structs ifnet, domain, protosw
and vnet_net, which requires the kernel and modules to be rebuilt, this
change should have no impact on nooptions VIMAGE builds, since vnet
destructors can only be called in VIMAGE kernels. Moreover,
destructor functions should be in general compiled in only in
options VIMAGE builds, except for kernel modules which can be safely
kldunloaded at run time.
Bump __FreeBSD_version to 800097.
Reviewed by: bz, julian
Approved by: rwatson, kib (re), julian (mentor)
2009-06-08 17:15:40 +00:00
|
|
|
#ifdef VIMAGE
|
2016-06-01 10:14:04 +00:00
|
|
|
static void
|
|
|
|
rip_destroy(void *unused __unused)
|
Introduce an infrastructure for dismantling vnet instances.
Vnet modules and protocol domains may now register destructor
functions to clean up and release per-module state. The destructor
mechanisms can be triggered by invoking "vimage -d", or a future
equivalent command which will be provided via the new jail framework.
While this patch introduces numerous placeholder destructor functions,
many of those are currently incomplete, thus leaking memory or (even
worse) failing to stop all running timers. Many of such issues are
already known and will be incrementaly fixed over the next weeks in
smaller incremental commits.
Apart from introducing new fields in structs ifnet, domain, protosw
and vnet_net, which requires the kernel and modules to be rebuilt, this
change should have no impact on nooptions VIMAGE builds, since vnet
destructors can only be called in VIMAGE kernels. Moreover,
destructor functions should be in general compiled in only in
options VIMAGE builds, except for kernel modules which can be safely
kldunloaded at run time.
Bump __FreeBSD_version to 800097.
Reviewed by: bz, julian
Approved by: rwatson, kib (re), julian (mentor)
2009-06-08 17:15:40 +00:00
|
|
|
{
|
|
|
|
|
2010-03-14 18:59:11 +00:00
|
|
|
in_pcbinfo_destroy(&V_ripcbinfo);
|
Introduce an infrastructure for dismantling vnet instances.
Vnet modules and protocol domains may now register destructor
functions to clean up and release per-module state. The destructor
mechanisms can be triggered by invoking "vimage -d", or a future
equivalent command which will be provided via the new jail framework.
While this patch introduces numerous placeholder destructor functions,
many of those are currently incomplete, thus leaking memory or (even
worse) failing to stop all running timers. Many of such issues are
already known and will be incrementaly fixed over the next weeks in
smaller incremental commits.
Apart from introducing new fields in structs ifnet, domain, protosw
and vnet_net, which requires the kernel and modules to be rebuilt, this
change should have no impact on nooptions VIMAGE builds, since vnet
destructors can only be called in VIMAGE kernels. Moreover,
destructor functions should be in general compiled in only in
options VIMAGE builds, except for kernel modules which can be safely
kldunloaded at run time.
Bump __FreeBSD_version to 800097.
Reviewed by: bz, julian
Approved by: rwatson, kib (re), julian (mentor)
2009-06-08 17:15:40 +00:00
|
|
|
}
|
2016-06-01 10:14:04 +00:00
|
|
|
VNET_SYSUNINIT(raw_ip, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, rip_destroy, NULL);
|
Introduce an infrastructure for dismantling vnet instances.
Vnet modules and protocol domains may now register destructor
functions to clean up and release per-module state. The destructor
mechanisms can be triggered by invoking "vimage -d", or a future
equivalent command which will be provided via the new jail framework.
While this patch introduces numerous placeholder destructor functions,
many of those are currently incomplete, thus leaking memory or (even
worse) failing to stop all running timers. Many of such issues are
already known and will be incrementaly fixed over the next weeks in
smaller incremental commits.
Apart from introducing new fields in structs ifnet, domain, protosw
and vnet_net, which requires the kernel and modules to be rebuilt, this
change should have no impact on nooptions VIMAGE builds, since vnet
destructors can only be called in VIMAGE kernels. Moreover,
destructor functions should be in general compiled in only in
options VIMAGE builds, except for kernel modules which can be safely
kldunloaded at run time.
Bump __FreeBSD_version to 800097.
Reviewed by: bz, julian
Approved by: rwatson, kib (re), julian (mentor)
2009-06-08 17:15:40 +00:00
|
|
|
#endif
|
|
|
|
|
2011-04-20 08:03:22 +00:00
|
|
|
#ifdef INET
|
2003-09-01 04:23:48 +00:00
|
|
|
static int
|
2008-07-18 10:47:07 +00:00
|
|
|
rip_append(struct inpcb *last, struct ip *ip, struct mbuf *n,
|
|
|
|
struct sockaddr_in *ripsrc)
|
2003-09-01 04:23:48 +00:00
|
|
|
{
|
|
|
|
int policyfail = 0;
|
|
|
|
|
Decompose the current single inpcbinfo lock into two locks:
- The existing ipi_lock continues to protect the global inpcb list and
inpcb counter. This lock is now relegated to a small number of
allocation and free operations, and occasional operations that walk
all connections (including, awkwardly, certain UDP multicast receive
operations -- something to revisit).
- A new ipi_hash_lock protects the two inpcbinfo hash tables for
looking up connections and bound sockets, manipulated using new
INP_HASH_*() macros. This lock, combined with inpcb locks, protects
the 4-tuple address space.
Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required. As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.
A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb. Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed. In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup. New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:
INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb
Callers must pass exactly one of these flags (for the time being).
Some notes:
- All protocols are updated to work within the new regime; especially,
TCP, UDPv4, and UDPv6. pcbinfo ipi_lock acquisitions are largely
eliminated, and global hash lock hold times are dramatically reduced
compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
is no longer available -- hash lookup locks are now held only very
briefly during inpcb lookup, rather than for potentially extended
periods. However, the pcbinfo ipi_lock will still be acquired if a
connection state might change such that a connection is added or
removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
callers to acquire hash locks and perform one or more lookups atomically
with 4-tuple allocation: this is required only for TCPv6, as there is no
in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
locking, which relates to source address selection. This needs
attention, as it likely significantly reduces parallelism in this code
for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
is no longer sufficient. A second check once the inpcb lock is held
should do the trick, keeping the general case from requiring the inpcb
lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
undesirable, and probably another argument is required to take care of
this (or a char array name field in the pcbinfo?).
This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics. It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.
Reviewed by: bz
Sponsored by: Juniper Networks, Inc.
2011-05-30 09:43:55 +00:00
|
|
|
INP_LOCK_ASSERT(last);
|
2004-05-04 00:10:16 +00:00
|
|
|
|
2017-02-06 08:49:57 +00:00
|
|
|
#if defined(IPSEC) || defined(IPSEC_SUPPORT)
|
2003-09-01 04:23:48 +00:00
|
|
|
/* check AH/ESP integrity. */
|
2017-02-06 08:49:57 +00:00
|
|
|
if (IPSEC_ENABLED(ipv4)) {
|
|
|
|
if (IPSEC_CHECK_POLICY(ipv4, n, last) != 0)
|
|
|
|
policyfail = 1;
|
2003-09-01 04:23:48 +00:00
|
|
|
}
|
2007-07-03 12:13:45 +00:00
|
|
|
#endif /* IPSEC */
|
2003-09-01 04:23:48 +00:00
|
|
|
#ifdef MAC
|
2007-10-24 19:04:04 +00:00
|
|
|
if (!policyfail && mac_inpcb_check_deliver(last, n) != 0)
|
2003-09-01 04:23:48 +00:00
|
|
|
policyfail = 1;
|
|
|
|
#endif
|
2005-08-22 16:13:08 +00:00
|
|
|
/* Check the minimum TTL for socket. */
|
|
|
|
if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl)
|
|
|
|
policyfail = 1;
|
2003-09-01 04:23:48 +00:00
|
|
|
if (!policyfail) {
|
|
|
|
struct mbuf *opts = NULL;
|
Reduce the number of unnecessary unlock-relocks on socket buffer mutexes
associated with performing a wakeup on the socket buffer:
- When performing an sbappend*() followed by a so[rw]wakeup(), explicitly
acquire the socket buffer lock and use the _locked() variants of both
calls. Note that the _locked() sowakeup() versions unlock the mutex on
return. This is done in uipc_send(), divert_packet(), mroute
socket_send(), raw_append(), tcp_reass(), tcp_input(), and udp_append().
- When the socket buffer lock is dropped before a sowakeup(), remove the
explicit unlock and use the _locked() sowakeup() variant. This is done
in soisdisconnecting(), soisdisconnected() when setting the can't send/
receive flags and dropping data, and in uipc_rcvd() which adjusting
back-pressure on the sockets.
For UNIX domain sockets running mpsafe with a contention-intensive SMP
mysql benchmark, this results in a 1.6% query rate improvement due to
reduce mutex costs.
2004-06-26 19:10:39 +00:00
|
|
|
struct socket *so;
|
2003-09-01 04:23:48 +00:00
|
|
|
|
Reduce the number of unnecessary unlock-relocks on socket buffer mutexes
associated with performing a wakeup on the socket buffer:
- When performing an sbappend*() followed by a so[rw]wakeup(), explicitly
acquire the socket buffer lock and use the _locked() variants of both
calls. Note that the _locked() sowakeup() versions unlock the mutex on
return. This is done in uipc_send(), divert_packet(), mroute
socket_send(), raw_append(), tcp_reass(), tcp_input(), and udp_append().
- When the socket buffer lock is dropped before a sowakeup(), remove the
explicit unlock and use the _locked() sowakeup() variant. This is done
in soisdisconnecting(), soisdisconnected() when setting the can't send/
receive flags and dropping data, and in uipc_rcvd() which adjusting
back-pressure on the sockets.
For UNIX domain sockets running mpsafe with a contention-intensive SMP
mysql benchmark, this results in a 1.6% query rate improvement due to
reduce mutex costs.
2004-06-26 19:10:39 +00:00
|
|
|
so = last->inp_socket;
|
2003-09-01 04:23:48 +00:00
|
|
|
if ((last->inp_flags & INP_CONTROLOPTS) ||
|
2005-10-04 18:19:21 +00:00
|
|
|
(so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
|
2003-09-01 04:23:48 +00:00
|
|
|
ip_savecontrol(last, &opts, ip, n);
|
Reduce the number of unnecessary unlock-relocks on socket buffer mutexes
associated with performing a wakeup on the socket buffer:
- When performing an sbappend*() followed by a so[rw]wakeup(), explicitly
acquire the socket buffer lock and use the _locked() variants of both
calls. Note that the _locked() sowakeup() versions unlock the mutex on
return. This is done in uipc_send(), divert_packet(), mroute
socket_send(), raw_append(), tcp_reass(), tcp_input(), and udp_append().
- When the socket buffer lock is dropped before a sowakeup(), remove the
explicit unlock and use the _locked() sowakeup() variant. This is done
in soisdisconnecting(), soisdisconnected() when setting the can't send/
receive flags and dropping data, and in uipc_rcvd() which adjusting
back-pressure on the sockets.
For UNIX domain sockets running mpsafe with a contention-intensive SMP
mysql benchmark, this results in a 1.6% query rate improvement due to
reduce mutex costs.
2004-06-26 19:10:39 +00:00
|
|
|
SOCKBUF_LOCK(&so->so_rcv);
|
|
|
|
if (sbappendaddr_locked(&so->so_rcv,
|
2008-07-18 10:47:07 +00:00
|
|
|
(struct sockaddr *)ripsrc, n, opts) == 0) {
|
2003-09-01 04:23:48 +00:00
|
|
|
/* should notify about lost packet */
|
|
|
|
m_freem(n);
|
|
|
|
if (opts)
|
|
|
|
m_freem(opts);
|
Reduce the number of unnecessary unlock-relocks on socket buffer mutexes
associated with performing a wakeup on the socket buffer:
- When performing an sbappend*() followed by a so[rw]wakeup(), explicitly
acquire the socket buffer lock and use the _locked() variants of both
calls. Note that the _locked() sowakeup() versions unlock the mutex on
return. This is done in uipc_send(), divert_packet(), mroute
socket_send(), raw_append(), tcp_reass(), tcp_input(), and udp_append().
- When the socket buffer lock is dropped before a sowakeup(), remove the
explicit unlock and use the _locked() sowakeup() variant. This is done
in soisdisconnecting(), soisdisconnected() when setting the can't send/
receive flags and dropping data, and in uipc_rcvd() which adjusting
back-pressure on the sockets.
For UNIX domain sockets running mpsafe with a contention-intensive SMP
mysql benchmark, this results in a 1.6% query rate improvement due to
reduce mutex costs.
2004-06-26 19:10:39 +00:00
|
|
|
SOCKBUF_UNLOCK(&so->so_rcv);
|
2003-09-01 04:23:48 +00:00
|
|
|
} else
|
Reduce the number of unnecessary unlock-relocks on socket buffer mutexes
associated with performing a wakeup on the socket buffer:
- When performing an sbappend*() followed by a so[rw]wakeup(), explicitly
acquire the socket buffer lock and use the _locked() variants of both
calls. Note that the _locked() sowakeup() versions unlock the mutex on
return. This is done in uipc_send(), divert_packet(), mroute
socket_send(), raw_append(), tcp_reass(), tcp_input(), and udp_append().
- When the socket buffer lock is dropped before a sowakeup(), remove the
explicit unlock and use the _locked() sowakeup() variant. This is done
in soisdisconnecting(), soisdisconnected() when setting the can't send/
receive flags and dropping data, and in uipc_rcvd() which adjusting
back-pressure on the sockets.
For UNIX domain sockets running mpsafe with a contention-intensive SMP
mysql benchmark, this results in a 1.6% query rate improvement due to
reduce mutex costs.
2004-06-26 19:10:39 +00:00
|
|
|
sorwakeup_locked(so);
|
2003-09-01 04:23:48 +00:00
|
|
|
} else
|
|
|
|
m_freem(n);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (policyfail);
|
2003-09-01 04:23:48 +00:00
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* Setup generic address and protocol structures for raw_input routine, then
|
|
|
|
* pass them along with mbuf chain.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
2014-08-08 01:57:15 +00:00
|
|
|
int
|
|
|
|
rip_input(struct mbuf **mp, int *offp, int proto)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2009-03-09 17:53:05 +00:00
|
|
|
struct ifnet *ifp;
|
2014-08-08 01:57:15 +00:00
|
|
|
struct mbuf *m = *mp;
|
2003-09-01 04:23:48 +00:00
|
|
|
struct ip *ip = mtod(m, struct ip *);
|
|
|
|
struct inpcb *inp, *last;
|
2008-07-18 10:47:07 +00:00
|
|
|
struct sockaddr_in ripsrc;
|
2008-07-26 17:32:15 +00:00
|
|
|
int hash;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2019-11-07 20:40:44 +00:00
|
|
|
NET_EPOCH_ASSERT();
|
|
|
|
|
2014-08-08 01:57:15 +00:00
|
|
|
*mp = NULL;
|
|
|
|
|
2008-07-18 10:47:07 +00:00
|
|
|
bzero(&ripsrc, sizeof(ripsrc));
|
|
|
|
ripsrc.sin_len = sizeof(ripsrc);
|
|
|
|
ripsrc.sin_family = AF_INET;
|
1994-05-24 10:09:53 +00:00
|
|
|
ripsrc.sin_addr = ip->ip_src;
|
2003-09-01 04:23:48 +00:00
|
|
|
last = NULL;
|
2009-03-09 17:53:05 +00:00
|
|
|
|
|
|
|
ifp = m->m_pkthdr.rcvif;
|
|
|
|
|
2008-07-26 17:32:15 +00:00
|
|
|
hash = INP_PCBHASH_RAW(proto, ip->ip_src.s_addr,
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
ip->ip_dst.s_addr, V_ripcbinfo.ipi_hashmask);
|
2018-06-12 22:18:20 +00:00
|
|
|
CK_LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[hash], inp_hash) {
|
2008-07-26 21:12:00 +00:00
|
|
|
if (inp->inp_ip_p != proto)
|
2008-07-26 17:32:15 +00:00
|
|
|
continue;
|
|
|
|
#ifdef INET6
|
2008-10-04 15:06:34 +00:00
|
|
|
/* XXX inp locking */
|
2008-07-26 17:32:15 +00:00
|
|
|
if ((inp->inp_vflag & INP_IPV4) == 0)
|
2008-07-26 21:12:00 +00:00
|
|
|
continue;
|
2008-07-26 17:32:15 +00:00
|
|
|
#endif
|
|
|
|
if (inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
|
2008-07-26 21:12:00 +00:00
|
|
|
continue;
|
2008-07-26 17:32:15 +00:00
|
|
|
if (inp->inp_faddr.s_addr != ip->ip_src.s_addr)
|
2008-07-26 21:12:00 +00:00
|
|
|
continue;
|
2008-12-16 03:18:59 +00:00
|
|
|
if (last != NULL) {
|
2008-07-26 17:32:15 +00:00
|
|
|
struct mbuf *n;
|
|
|
|
|
2016-09-15 07:41:48 +00:00
|
|
|
n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
|
2008-07-26 17:32:15 +00:00
|
|
|
if (n != NULL)
|
2018-06-21 20:18:23 +00:00
|
|
|
(void) rip_append(last, ip, n, &ripsrc);
|
2008-07-26 17:32:15 +00:00
|
|
|
/* XXX count dropped packet */
|
|
|
|
INP_RUNLOCK(last);
|
2018-06-21 20:18:23 +00:00
|
|
|
last = NULL;
|
2008-07-26 17:32:15 +00:00
|
|
|
}
|
2008-10-04 15:06:34 +00:00
|
|
|
INP_RLOCK(inp);
|
2018-06-21 20:18:23 +00:00
|
|
|
if (__predict_false(inp->inp_flags2 & INP_FREED))
|
|
|
|
goto skip_1;
|
|
|
|
if (jailed_without_vnet(inp->inp_cred)) {
|
|
|
|
/*
|
|
|
|
* XXX: If faddr was bound to multicast group,
|
|
|
|
* jailed raw socket will drop datagram.
|
|
|
|
*/
|
|
|
|
if (prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0)
|
|
|
|
goto skip_1;
|
2018-06-21 17:24:10 +00:00
|
|
|
}
|
2018-06-21 20:18:23 +00:00
|
|
|
last = inp;
|
|
|
|
continue;
|
|
|
|
skip_1:
|
|
|
|
INP_RUNLOCK(inp);
|
2008-07-26 17:32:15 +00:00
|
|
|
}
|
2018-06-12 22:18:20 +00:00
|
|
|
CK_LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[0], inp_hash) {
|
2008-07-26 21:12:00 +00:00
|
|
|
if (inp->inp_ip_p && inp->inp_ip_p != proto)
|
2003-09-01 04:23:48 +00:00
|
|
|
continue;
|
1999-12-22 19:13:38 +00:00
|
|
|
#ifdef INET6
|
2008-10-04 15:06:34 +00:00
|
|
|
/* XXX inp locking */
|
1999-12-22 19:13:38 +00:00
|
|
|
if ((inp->inp_vflag & INP_IPV4) == 0)
|
2008-07-26 21:12:00 +00:00
|
|
|
continue;
|
1999-12-22 19:13:38 +00:00
|
|
|
#endif
|
2009-03-09 17:53:05 +00:00
|
|
|
if (!in_nullhost(inp->inp_laddr) &&
|
|
|
|
!in_hosteq(inp->inp_laddr, ip->ip_dst))
|
2008-07-26 21:12:00 +00:00
|
|
|
continue;
|
2009-03-09 17:53:05 +00:00
|
|
|
if (!in_nullhost(inp->inp_faddr) &&
|
|
|
|
!in_hosteq(inp->inp_faddr, ip->ip_src))
|
2009-02-05 14:06:09 +00:00
|
|
|
continue;
|
2018-06-21 20:18:23 +00:00
|
|
|
if (last != NULL) {
|
|
|
|
struct mbuf *n;
|
|
|
|
|
|
|
|
n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
|
|
|
|
if (n != NULL)
|
|
|
|
(void) rip_append(last, ip, n, &ripsrc);
|
|
|
|
/* XXX count dropped packet */
|
|
|
|
INP_RUNLOCK(last);
|
|
|
|
last = NULL;
|
|
|
|
}
|
|
|
|
INP_RLOCK(inp);
|
|
|
|
if (__predict_false(inp->inp_flags2 & INP_FREED))
|
|
|
|
goto skip_2;
|
2009-12-13 13:57:32 +00:00
|
|
|
if (jailed_without_vnet(inp->inp_cred)) {
|
2009-03-09 17:53:05 +00:00
|
|
|
/*
|
|
|
|
* Allow raw socket in jail to receive multicast;
|
|
|
|
* assume process had PRIV_NETINET_RAW at attach,
|
|
|
|
* and fall through into normal filter path if so.
|
|
|
|
*/
|
|
|
|
if (!IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) &&
|
|
|
|
prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0)
|
2018-06-21 20:18:23 +00:00
|
|
|
goto skip_2;
|
2009-03-09 17:53:05 +00:00
|
|
|
}
|
|
|
|
/*
|
|
|
|
* If this raw socket has multicast state, and we
|
|
|
|
* have received a multicast, check if this socket
|
|
|
|
* should receive it, as multicast filtering is now
|
|
|
|
* the responsibility of the transport layer.
|
|
|
|
*/
|
|
|
|
if (inp->inp_moptions != NULL &&
|
|
|
|
IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
|
2009-11-15 11:07:22 +00:00
|
|
|
/*
|
|
|
|
* If the incoming datagram is for IGMP, allow it
|
|
|
|
* through unconditionally to the raw socket.
|
|
|
|
*
|
|
|
|
* In the case of IGMPv2, we may not have explicitly
|
|
|
|
* joined the group, and may have set IFF_ALLMULTI
|
|
|
|
* on the interface. imo_multi_filter() may discard
|
|
|
|
* control traffic we actually need to see.
|
|
|
|
*
|
|
|
|
* Userland multicast routing daemons should continue
|
|
|
|
* filter the control traffic appropriately.
|
|
|
|
*/
|
2009-03-09 17:53:05 +00:00
|
|
|
int blocked;
|
|
|
|
|
2009-11-15 11:07:22 +00:00
|
|
|
blocked = MCAST_PASS;
|
|
|
|
if (proto != IPPROTO_IGMP) {
|
|
|
|
struct sockaddr_in group;
|
|
|
|
|
|
|
|
bzero(&group, sizeof(struct sockaddr_in));
|
|
|
|
group.sin_len = sizeof(struct sockaddr_in);
|
|
|
|
group.sin_family = AF_INET;
|
|
|
|
group.sin_addr = ip->ip_dst;
|
|
|
|
|
|
|
|
blocked = imo_multi_filter(inp->inp_moptions,
|
|
|
|
ifp,
|
|
|
|
(struct sockaddr *)&group,
|
|
|
|
(struct sockaddr *)&ripsrc);
|
|
|
|
}
|
2009-03-09 17:53:05 +00:00
|
|
|
|
|
|
|
if (blocked != MCAST_PASS) {
|
2009-04-11 23:35:20 +00:00
|
|
|
IPSTAT_INC(ips_notmember);
|
2018-06-21 20:18:23 +00:00
|
|
|
goto skip_2;
|
2009-03-09 17:53:05 +00:00
|
|
|
}
|
|
|
|
}
|
1996-11-11 04:56:32 +00:00
|
|
|
last = inp;
|
2018-06-21 20:18:23 +00:00
|
|
|
continue;
|
|
|
|
skip_2:
|
|
|
|
INP_RUNLOCK(inp);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2003-09-01 04:23:48 +00:00
|
|
|
if (last != NULL) {
|
2008-07-18 10:47:07 +00:00
|
|
|
if (rip_append(last, ip, m, &ripsrc) != 0)
|
2009-04-11 23:35:20 +00:00
|
|
|
IPSTAT_INC(ips_delivered);
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RUNLOCK(last);
|
1994-05-24 10:09:53 +00:00
|
|
|
} else {
|
2016-05-25 13:48:26 +00:00
|
|
|
if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) {
|
|
|
|
IPSTAT_INC(ips_noproto);
|
|
|
|
IPSTAT_DEC(ips_delivered);
|
2016-05-25 15:54:21 +00:00
|
|
|
icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL, 0, 0);
|
|
|
|
} else {
|
|
|
|
m_freem(m);
|
2016-05-25 13:48:26 +00:00
|
|
|
}
|
2001-06-11 12:39:29 +00:00
|
|
|
}
|
2014-08-08 01:57:15 +00:00
|
|
|
return (IPPROTO_DONE);
|
1995-03-16 16:25:55 +00:00
|
|
|
}
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* Generate IP header and pass packet to ip_output. Tack on options user may
|
|
|
|
* have setup with control call.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
|
|
|
int
|
2014-08-15 02:43:02 +00:00
|
|
|
rip_output(struct mbuf *m, struct socket *so, ...)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
2003-09-01 04:23:48 +00:00
|
|
|
struct ip *ip;
|
2004-06-03 03:15:29 +00:00
|
|
|
int error;
|
2003-09-01 04:23:48 +00:00
|
|
|
struct inpcb *inp = sotoinpcb(so);
|
2014-08-15 02:43:02 +00:00
|
|
|
va_list ap;
|
|
|
|
u_long dst;
|
2004-09-05 02:34:12 +00:00
|
|
|
int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
|
|
|
|
IP_ALLOWBROADCAST;
|
2019-04-13 10:47:47 +00:00
|
|
|
int cnt, hlen;
|
2019-02-12 10:17:21 +00:00
|
|
|
u_char opttype, optlen, *cp;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2014-08-15 02:43:02 +00:00
|
|
|
va_start(ap, so);
|
|
|
|
dst = va_arg(ap, u_long);
|
|
|
|
va_end(ap);
|
|
|
|
|
1994-05-24 10:09:53 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* If the user handed us a complete IP packet, use it. Otherwise,
|
|
|
|
* allocate an mbuf for a header and fill it in.
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
|
|
|
if ((inp->inp_flags & INP_HDRINCL) == 0) {
|
1996-10-25 17:57:53 +00:00
|
|
|
if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
|
|
|
|
m_freem(m);
|
|
|
|
return(EMSGSIZE);
|
|
|
|
}
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, sizeof(struct ip), M_NOWAIT);
|
2003-08-26 14:11:48 +00:00
|
|
|
if (m == NULL)
|
|
|
|
return(ENOBUFS);
|
2004-06-03 03:15:29 +00:00
|
|
|
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RLOCK(inp);
|
1994-05-24 10:09:53 +00:00
|
|
|
ip = mtod(m, struct ip *);
|
2001-03-09 12:22:51 +00:00
|
|
|
ip->ip_tos = inp->inp_ip_tos;
|
2005-09-26 20:25:16 +00:00
|
|
|
if (inp->inp_flags & INP_DONTFRAG)
|
2012-10-22 21:09:03 +00:00
|
|
|
ip->ip_off = htons(IP_DF);
|
2005-09-26 20:25:16 +00:00
|
|
|
else
|
2012-10-22 21:09:03 +00:00
|
|
|
ip->ip_off = htons(0);
|
1997-04-03 05:14:45 +00:00
|
|
|
ip->ip_p = inp->inp_ip_p;
|
2012-10-22 21:09:03 +00:00
|
|
|
ip->ip_len = htons(m->m_pkthdr.len);
|
2009-02-05 14:06:09 +00:00
|
|
|
ip->ip_src = inp->inp_laddr;
|
2014-04-24 12:52:31 +00:00
|
|
|
ip->ip_dst.s_addr = dst;
|
2010-04-27 15:07:08 +00:00
|
|
|
if (jailed(inp->inp_cred)) {
|
|
|
|
/*
|
|
|
|
* prison_local_ip4() would be good enough but would
|
|
|
|
* let a source of INADDR_ANY pass, which we do not
|
2014-04-24 12:52:31 +00:00
|
|
|
* want to see from jails.
|
2010-04-27 15:07:08 +00:00
|
|
|
*/
|
2014-04-24 12:52:31 +00:00
|
|
|
if (ip->ip_src.s_addr == INADDR_ANY) {
|
|
|
|
error = in_pcbladdr(inp, &ip->ip_dst, &ip->ip_src,
|
|
|
|
inp->inp_cred);
|
|
|
|
} else {
|
2010-04-27 15:07:08 +00:00
|
|
|
error = prison_local_ip4(inp->inp_cred,
|
|
|
|
&ip->ip_src);
|
2014-04-24 12:52:31 +00:00
|
|
|
}
|
2010-04-27 15:07:08 +00:00
|
|
|
if (error != 0) {
|
|
|
|
INP_RUNLOCK(inp);
|
|
|
|
m_freem(m);
|
|
|
|
return (error);
|
|
|
|
}
|
MFp4:
Bring in updated jail support from bz_jail branch.
This enhances the current jail implementation to permit multiple
addresses per jail. In addtion to IPv4, IPv6 is supported as well.
Due to updated checks it is even possible to have jails without
an IP address at all, which basically gives one a chroot with
restricted process view, no networking,..
SCTP support was updated and supports IPv6 in jails as well.
Cpuset support permits jails to be bound to specific processor
sets after creation.
Jails can have an unrestricted (no duplicate protection, etc.) name
in addition to the hostname. The jail name cannot be changed from
within a jail and is considered to be used for management purposes
or as audit-token in the future.
DDB 'show jails' command was added to aid debugging.
Proper compat support permits 32bit jail binaries to be used on 64bit
systems to manage jails. Also backward compatibility was preserved where
possible: for jail v1 syscalls, as well as with user space management
utilities.
Both jail as well as prison version were updated for the new features.
A gap was intentionally left as the intermediate versions had been
used by various patches floating around the last years.
Bump __FreeBSD_version for the afore mentioned and in kernel changes.
Special thanks to:
- Pawel Jakub Dawidek (pjd) for his multi-IPv4 patches
and Olivier Houchard (cognet) for initial single-IPv6 patches.
- Jeff Roberson (jeff) and Randall Stewart (rrs) for their
help, ideas and review on cpuset and SCTP support.
- Robert Watson (rwatson) for lots and lots of help, discussions,
suggestions and review of most of the patch at various stages.
- John Baldwin (jhb) for his help.
- Simon L. Nielsen (simon) as early adopter testing changes
on cluster machines as well as all the testers and people
who provided feedback the last months on freebsd-jail and
other channels.
- My employer, CK Software GmbH, for the support so I could work on this.
Reviewed by: (see above)
MFC after: 3 months (this is just so that I get the mail)
X-MFC Before: 7.2-RELEASE if possible
2008-11-29 14:32:14 +00:00
|
|
|
}
|
2001-03-09 12:22:51 +00:00
|
|
|
ip->ip_ttl = inp->inp_ip_ttl;
|
1994-05-24 10:09:53 +00:00
|
|
|
} else {
|
1996-10-25 17:57:53 +00:00
|
|
|
if (m->m_pkthdr.len > IP_MAXPACKET) {
|
|
|
|
m_freem(m);
|
|
|
|
return(EMSGSIZE);
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
ip = mtod(m, struct ip *);
|
2019-04-13 10:47:47 +00:00
|
|
|
hlen = ip->ip_hl << 2;
|
|
|
|
if (m->m_len < hlen) {
|
|
|
|
m = m_pullup(m, hlen);
|
|
|
|
if (m == NULL)
|
|
|
|
return (EINVAL);
|
|
|
|
ip = mtod(m, struct ip *);
|
2004-04-26 19:46:52 +00:00
|
|
|
}
|
2008-07-05 18:03:39 +00:00
|
|
|
|
2019-04-13 10:47:47 +00:00
|
|
|
INP_RLOCK(inp);
|
2008-07-05 18:03:39 +00:00
|
|
|
/*
|
|
|
|
* Don't allow both user specified and setsockopt options,
|
|
|
|
* and don't allow packet length sizes that will crash.
|
|
|
|
*/
|
2019-04-13 10:47:47 +00:00
|
|
|
if ((hlen < sizeof (*ip))
|
|
|
|
|| ((hlen > sizeof (*ip)) && inp->inp_options)
|
|
|
|
|| (ntohs(ip->ip_len) != m->m_pkthdr.len)) {
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RUNLOCK(inp);
|
1996-03-13 08:02:45 +00:00
|
|
|
m_freem(m);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EINVAL);
|
1996-03-13 08:02:45 +00:00
|
|
|
}
|
2019-04-13 10:47:47 +00:00
|
|
|
error = prison_check_ip4(inp->inp_cred, &ip->ip_src);
|
|
|
|
if (error != 0) {
|
|
|
|
INP_RUNLOCK(inp);
|
|
|
|
m_freem(m);
|
|
|
|
return (error);
|
|
|
|
}
|
2019-02-12 10:17:21 +00:00
|
|
|
/*
|
|
|
|
* Don't allow IP options which do not have the required
|
|
|
|
* structure as specified in section 3.1 of RFC 791 on
|
|
|
|
* pages 15-23.
|
|
|
|
*/
|
|
|
|
cp = (u_char *)(ip + 1);
|
2019-04-13 10:47:47 +00:00
|
|
|
cnt = hlen - sizeof (struct ip);
|
2019-02-12 10:17:21 +00:00
|
|
|
for (; cnt > 0; cnt -= optlen, cp += optlen) {
|
|
|
|
opttype = cp[IPOPT_OPTVAL];
|
|
|
|
if (opttype == IPOPT_EOL)
|
|
|
|
break;
|
|
|
|
if (opttype == IPOPT_NOP) {
|
|
|
|
optlen = 1;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (cnt < IPOPT_OLEN + sizeof(u_char)) {
|
|
|
|
INP_RUNLOCK(inp);
|
|
|
|
m_freem(m);
|
|
|
|
return (EINVAL);
|
|
|
|
}
|
|
|
|
optlen = cp[IPOPT_OLEN];
|
|
|
|
if (optlen < IPOPT_OLEN + sizeof(u_char) ||
|
|
|
|
optlen > cnt) {
|
|
|
|
INP_RUNLOCK(inp);
|
|
|
|
m_freem(m);
|
|
|
|
return (EINVAL);
|
|
|
|
}
|
|
|
|
}
|
2015-04-01 22:26:39 +00:00
|
|
|
/*
|
|
|
|
* This doesn't allow application to specify ID of zero,
|
|
|
|
* but we got this limitation from the beginning of history.
|
|
|
|
*/
|
1994-05-24 10:09:53 +00:00
|
|
|
if (ip->ip_id == 0)
|
2015-04-01 22:26:39 +00:00
|
|
|
ip_fillid(ip);
|
2008-07-05 18:03:39 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* XXX prevent ip_output from overwriting header fields.
|
|
|
|
*/
|
1994-05-24 10:09:53 +00:00
|
|
|
flags |= IP_RAWOUTPUT;
|
2009-04-11 23:35:20 +00:00
|
|
|
IPSTAT_INC(ips_rawout);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
1999-12-22 19:13:38 +00:00
|
|
|
|
2006-09-06 19:04:36 +00:00
|
|
|
if (inp->inp_flags & INP_ONESBCAST)
|
2003-08-20 14:46:40 +00:00
|
|
|
flags |= IP_SENDONES;
|
|
|
|
|
2004-06-03 03:15:29 +00:00
|
|
|
#ifdef MAC
|
2007-10-24 19:04:04 +00:00
|
|
|
mac_inpcb_create_mbuf(inp, m);
|
2004-06-03 03:15:29 +00:00
|
|
|
#endif
|
|
|
|
|
|
|
|
error = ip_output(m, inp->inp_options, NULL, flags,
|
|
|
|
inp->inp_moptions, inp);
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RUNLOCK(inp);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Raw IP socket option processing.
|
2003-07-18 16:10:36 +00:00
|
|
|
*
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
* IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
|
|
|
|
* only be created by a privileged process, and as such, socket option
|
|
|
|
* operations to manage system properties on any raw socket were allowed to
|
|
|
|
* take place without explicit additional access control checks. However,
|
|
|
|
* raw sockets can now also be created in jail(), and therefore explicit
|
|
|
|
* checks are now required. Likewise, raw sockets can be used by a process
|
|
|
|
* after it gives up privilege, so some caution is required. For options
|
|
|
|
* passed down to the IP layer via ip_ctloutput(), checks are assumed to be
|
|
|
|
* performed in ip_ctloutput() and therefore no check occurs here.
|
2007-07-02 15:44:30 +00:00
|
|
|
* Unilaterally checking priv_check() here breaks normal IP socket option
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
* operations on raw sockets.
|
|
|
|
*
|
|
|
|
* When adding new socket options here, make sure to add access control
|
|
|
|
* checks here as necessary.
|
2012-01-02 09:18:58 +00:00
|
|
|
*
|
|
|
|
* XXX-BZ inp locking?
|
1994-05-24 10:09:53 +00:00
|
|
|
*/
|
|
|
|
int
|
2003-09-01 04:23:48 +00:00
|
|
|
rip_ctloutput(struct socket *so, struct sockopt *sopt)
|
1994-05-24 10:09:53 +00:00
|
|
|
{
|
1998-08-23 03:07:17 +00:00
|
|
|
struct inpcb *inp = sotoinpcb(so);
|
|
|
|
int error, optval;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2008-11-19 19:19:30 +00:00
|
|
|
if (sopt->sopt_level != IPPROTO_IP) {
|
|
|
|
if ((sopt->sopt_level == SOL_SOCKET) &&
|
|
|
|
(sopt->sopt_name == SO_SETFIB)) {
|
|
|
|
inp->inp_inc.inc_fibnum = so->so_fibnum;
|
|
|
|
return (0);
|
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
return (EINVAL);
|
2008-11-19 19:19:30 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1998-08-23 03:07:17 +00:00
|
|
|
error = 0;
|
|
|
|
switch (sopt->sopt_dir) {
|
|
|
|
case SOPT_GET:
|
|
|
|
switch (sopt->sopt_name) {
|
|
|
|
case IP_HDRINCL:
|
|
|
|
optval = inp->inp_flags & INP_HDRINCL;
|
|
|
|
error = sooptcopyout(sopt, &optval, sizeof optval);
|
|
|
|
break;
|
|
|
|
|
2009-12-02 15:50:43 +00:00
|
|
|
case IP_FW3: /* generic ipfw v.3 functions */
|
2001-11-04 22:56:25 +00:00
|
|
|
case IP_FW_ADD: /* ADD actually returns the body... */
|
1998-08-23 03:07:17 +00:00
|
|
|
case IP_FW_GET:
|
2004-06-09 20:10:38 +00:00
|
|
|
case IP_FW_TABLE_GETSIZE:
|
|
|
|
case IP_FW_TABLE_LIST:
|
2006-12-29 21:59:17 +00:00
|
|
|
case IP_FW_NAT_GET_CONFIG:
|
|
|
|
case IP_FW_NAT_GET_LOG:
|
2009-10-11 05:59:43 +00:00
|
|
|
if (V_ip_fw_ctl_ptr != NULL)
|
|
|
|
error = V_ip_fw_ctl_ptr(sopt);
|
2001-11-04 22:56:25 +00:00
|
|
|
else
|
|
|
|
error = ENOPROTOOPT;
|
1998-08-23 03:07:17 +00:00
|
|
|
break;
|
|
|
|
|
2009-12-02 15:50:43 +00:00
|
|
|
case IP_DUMMYNET3: /* generic dummynet v.3 functions */
|
1998-12-14 18:09:13 +00:00
|
|
|
case IP_DUMMYNET_GET:
|
2004-08-17 22:05:54 +00:00
|
|
|
if (ip_dn_ctl_ptr != NULL)
|
1998-12-14 18:09:13 +00:00
|
|
|
error = ip_dn_ctl_ptr(sopt);
|
2001-11-04 22:56:25 +00:00
|
|
|
else
|
|
|
|
error = ENOPROTOOPT;
|
1998-12-14 18:09:13 +00:00
|
|
|
break ;
|
1998-08-23 03:07:17 +00:00
|
|
|
|
|
|
|
case MRT_INIT:
|
|
|
|
case MRT_DONE:
|
|
|
|
case MRT_ADD_VIF:
|
|
|
|
case MRT_DEL_VIF:
|
|
|
|
case MRT_ADD_MFC:
|
|
|
|
case MRT_DEL_MFC:
|
|
|
|
case MRT_VERSION:
|
|
|
|
case MRT_ASSERT:
|
2003-08-07 18:16:59 +00:00
|
|
|
case MRT_API_SUPPORT:
|
|
|
|
case MRT_API_CONFIG:
|
|
|
|
case MRT_ADD_BW_UPCALL:
|
|
|
|
case MRT_DEL_BW_UPCALL:
|
2006-11-06 13:42:10 +00:00
|
|
|
error = priv_check(curthread, PRIV_NETINET_MROUTE);
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
if (error != 0)
|
|
|
|
return (error);
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
|
|
|
|
EOPNOTSUPP;
|
1998-08-23 03:07:17 +00:00
|
|
|
break;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1998-08-23 03:07:17 +00:00
|
|
|
default:
|
|
|
|
error = ip_ctloutput(so, sopt);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
|
|
|
|
case SOPT_SET:
|
|
|
|
switch (sopt->sopt_name) {
|
|
|
|
case IP_HDRINCL:
|
|
|
|
error = sooptcopyin(sopt, &optval, sizeof optval,
|
|
|
|
sizeof optval);
|
|
|
|
if (error)
|
|
|
|
break;
|
|
|
|
if (optval)
|
1995-09-21 19:59:43 +00:00
|
|
|
inp->inp_flags |= INP_HDRINCL;
|
|
|
|
else
|
|
|
|
inp->inp_flags &= ~INP_HDRINCL;
|
1998-08-23 03:07:17 +00:00
|
|
|
break;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2009-12-02 15:50:43 +00:00
|
|
|
case IP_FW3: /* generic ipfw v.3 functions */
|
2001-11-26 10:05:58 +00:00
|
|
|
case IP_FW_ADD:
|
1998-08-23 03:07:17 +00:00
|
|
|
case IP_FW_DEL:
|
|
|
|
case IP_FW_FLUSH:
|
|
|
|
case IP_FW_ZERO:
|
1999-08-01 16:57:24 +00:00
|
|
|
case IP_FW_RESETLOG:
|
2004-06-09 20:10:38 +00:00
|
|
|
case IP_FW_TABLE_ADD:
|
|
|
|
case IP_FW_TABLE_DEL:
|
|
|
|
case IP_FW_TABLE_FLUSH:
|
2006-12-29 21:59:17 +00:00
|
|
|
case IP_FW_NAT_CFG:
|
|
|
|
case IP_FW_NAT_DEL:
|
2009-10-11 05:59:43 +00:00
|
|
|
if (V_ip_fw_ctl_ptr != NULL)
|
|
|
|
error = V_ip_fw_ctl_ptr(sopt);
|
2001-11-04 22:56:25 +00:00
|
|
|
else
|
|
|
|
error = ENOPROTOOPT;
|
1998-08-23 03:07:17 +00:00
|
|
|
break;
|
1995-01-12 13:06:32 +00:00
|
|
|
|
2009-12-02 15:50:43 +00:00
|
|
|
case IP_DUMMYNET3: /* generic dummynet v.3 functions */
|
1998-12-14 18:09:13 +00:00
|
|
|
case IP_DUMMYNET_CONFIGURE:
|
|
|
|
case IP_DUMMYNET_DEL:
|
|
|
|
case IP_DUMMYNET_FLUSH:
|
2004-08-17 22:05:54 +00:00
|
|
|
if (ip_dn_ctl_ptr != NULL)
|
1998-12-14 18:09:13 +00:00
|
|
|
error = ip_dn_ctl_ptr(sopt);
|
2001-11-04 22:56:25 +00:00
|
|
|
else
|
|
|
|
error = ENOPROTOOPT ;
|
1998-12-14 18:09:13 +00:00
|
|
|
break ;
|
1996-08-21 21:37:07 +00:00
|
|
|
|
1998-08-23 03:07:17 +00:00
|
|
|
case IP_RSVP_ON:
|
2006-11-06 13:42:10 +00:00
|
|
|
error = priv_check(curthread, PRIV_NETINET_MROUTE);
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
if (error != 0)
|
|
|
|
return (error);
|
1998-08-23 03:07:17 +00:00
|
|
|
error = ip_rsvp_init(so);
|
|
|
|
break;
|
1994-09-06 22:42:31 +00:00
|
|
|
|
1998-08-23 03:07:17 +00:00
|
|
|
case IP_RSVP_OFF:
|
2006-11-06 13:42:10 +00:00
|
|
|
error = priv_check(curthread, PRIV_NETINET_MROUTE);
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
if (error != 0)
|
|
|
|
return (error);
|
1998-08-23 03:07:17 +00:00
|
|
|
error = ip_rsvp_done();
|
|
|
|
break;
|
1994-09-06 22:42:31 +00:00
|
|
|
|
1998-08-23 03:07:17 +00:00
|
|
|
case IP_RSVP_VIF_ON:
|
|
|
|
case IP_RSVP_VIF_OFF:
|
2006-11-06 13:42:10 +00:00
|
|
|
error = priv_check(curthread, PRIV_NETINET_MROUTE);
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
if (error != 0)
|
|
|
|
return (error);
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
error = ip_rsvp_vif ?
|
|
|
|
ip_rsvp_vif(so, sopt) : EINVAL;
|
1998-08-23 03:07:17 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case MRT_INIT:
|
|
|
|
case MRT_DONE:
|
|
|
|
case MRT_ADD_VIF:
|
|
|
|
case MRT_DEL_VIF:
|
|
|
|
case MRT_ADD_MFC:
|
|
|
|
case MRT_DEL_MFC:
|
|
|
|
case MRT_VERSION:
|
|
|
|
case MRT_ASSERT:
|
2003-08-07 18:16:59 +00:00
|
|
|
case MRT_API_SUPPORT:
|
|
|
|
case MRT_API_CONFIG:
|
|
|
|
case MRT_ADD_BW_UPCALL:
|
|
|
|
case MRT_DEL_BW_UPCALL:
|
2006-11-06 13:42:10 +00:00
|
|
|
error = priv_check(curthread, PRIV_NETINET_MROUTE);
|
When the access control on creating raw sockets was modified so that
processes in jail could create raw sockets, additional access control
checks were added to raw IP sockets to limit the ways in which those
sockets could be used. Specifically, only the socket option IP_HDRINCL
was permitted in rip_ctloutput(). Other socket options were protected
by a call to suser(). This change was required to prevent processes
in a Jail from modifying system properties such as multicast routing
and firewall rule sets.
However, it also introduced a regression: processes that create a raw
socket with root privilege, but then downgraded credential (i.e., a
daemon giving up root, or a setuid process switching back to the real
uid) could no longer issue other unprivileged generic IP socket option
operations, such as IP_TOS, IP_TTL, and the multicast group membership
options, which prevented multicast routing daemons (and some other
tools) from operating correctly.
This change pushes the access control decision down to the granularity
of individual socket options, rather than all socket options, on raw
IP sockets. When rip_ctloutput() doesn't implement an option, it will
now pass the request directly to in_control() without an access
control check. This should restore the functionality of the generic
IP socket options for raw sockets in the above-described scenarios,
which may be confirmed with the ipsockopt regression test.
RELENG_5 candidate.
Reviewed by: csjp
2004-10-12 16:47:25 +00:00
|
|
|
if (error != 0)
|
|
|
|
return (error);
|
Massive cleanup of the ip_mroute code.
No functional changes, but:
+ the mrouting module now should behave the same as the compiled-in
version (it did not before, some of the rsvp code was not loaded
properly);
+ netinet/ip_mroute.c is now truly optional;
+ removed some redundant/unused code;
+ changed many instances of '0' to NULL and INADDR_ANY as appropriate;
+ removed several static variables to make the code more SMP-friendly;
+ fixed some minor bugs in the mrouting code (mostly, incorrect return
values from functions).
This commit is also a prerequisite to the addition of support for PIM,
which i would like to put in before DP2 (it does not change any of
the existing APIs, anyways).
Note, in the process we found out that some device drivers fail to
properly handle changes in IFF_ALLMULTI, leading to interesting
behaviour when a multicast router is started. This bug is not
corrected by this commit, and will be fixed with a separate commit.
Detailed changes:
--------------------
netinet/ip_mroute.c all the above.
conf/files make ip_mroute.c optional
net/route.c fix mrt_ioctl hook
netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here
together with other rsvp code, and a couple
of indentation fixes.
netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks
netinet/ip_var.h rsvp function hooks
netinet/raw_ip.c hooks for mrouting and rsvp functions, plus
interface cleanup.
netinet/ip_mroute.h remove an unused and optional field from a struct
Most of the code is from Pavlin Radoslavov and the XORP project
Reviewed by: sam
MFC after: 1 week
2002-11-15 22:53:53 +00:00
|
|
|
error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
|
|
|
|
EOPNOTSUPP;
|
1998-08-23 03:07:17 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
|
|
|
error = ip_ctloutput(so, sopt);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
break;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
1998-08-23 03:07:17 +00:00
|
|
|
|
|
|
|
return (error);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
|
|
|
|
1997-02-13 19:46:45 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* This function exists solely to receive the PRC_IFDOWN messages which are
|
|
|
|
* sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, and calls
|
|
|
|
* in_ifadown() to remove all routes corresponding to that address. It also
|
|
|
|
* receives the PRC_IFUP messages from if_up() and reinstalls the interface
|
|
|
|
* routes.
|
1997-02-13 19:46:45 +00:00
|
|
|
*/
|
|
|
|
void
|
2003-09-01 04:23:48 +00:00
|
|
|
rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
|
1997-02-13 19:46:45 +00:00
|
|
|
{
|
2015-07-29 08:12:05 +00:00
|
|
|
struct rm_priotracker in_ifa_tracker;
|
1997-02-13 19:46:45 +00:00
|
|
|
struct in_ifaddr *ia;
|
|
|
|
struct ifnet *ifp;
|
|
|
|
int err;
|
|
|
|
int flags;
|
|
|
|
|
1998-08-23 03:07:17 +00:00
|
|
|
switch (cmd) {
|
1997-02-13 19:46:45 +00:00
|
|
|
case PRC_IFDOWN:
|
2015-07-29 08:12:05 +00:00
|
|
|
IN_IFADDR_RLOCK(&in_ifa_tracker);
|
2018-05-18 20:13:34 +00:00
|
|
|
CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
|
1997-02-13 19:46:45 +00:00
|
|
|
if (ia->ia_ifa.ifa_addr == sa
|
|
|
|
&& (ia->ia_flags & IFA_ROUTE)) {
|
2009-06-25 11:52:33 +00:00
|
|
|
ifa_ref(&ia->ia_ifa);
|
2015-07-29 08:12:05 +00:00
|
|
|
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
|
1997-02-13 19:46:45 +00:00
|
|
|
/*
|
2013-11-01 10:18:41 +00:00
|
|
|
* in_scrubprefix() kills the interface route.
|
1997-02-13 19:46:45 +00:00
|
|
|
*/
|
2013-11-01 10:18:41 +00:00
|
|
|
in_scrubprefix(ia, 0);
|
1997-02-13 19:46:45 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* in_ifadown gets rid of all the rest of the
|
|
|
|
* routes. This is not quite the right thing
|
|
|
|
* to do, but at least if we are running a
|
|
|
|
* routing process they will come back.
|
1997-02-13 19:46:45 +00:00
|
|
|
*/
|
2001-05-11 14:37:34 +00:00
|
|
|
in_ifadown(&ia->ia_ifa, 0);
|
2009-06-25 11:52:33 +00:00
|
|
|
ifa_free(&ia->ia_ifa);
|
1997-02-13 19:46:45 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
2009-06-25 11:52:33 +00:00
|
|
|
if (ia == NULL) /* If ia matched, already unlocked. */
|
2015-07-29 08:12:05 +00:00
|
|
|
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
|
1997-02-13 19:46:45 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case PRC_IFUP:
|
2015-07-29 08:12:05 +00:00
|
|
|
IN_IFADDR_RLOCK(&in_ifa_tracker);
|
2018-05-18 20:13:34 +00:00
|
|
|
CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
|
1997-02-13 19:46:45 +00:00
|
|
|
if (ia->ia_ifa.ifa_addr == sa)
|
|
|
|
break;
|
|
|
|
}
|
2009-06-25 11:52:33 +00:00
|
|
|
if (ia == NULL || (ia->ia_flags & IFA_ROUTE)) {
|
2015-07-29 08:12:05 +00:00
|
|
|
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
|
1997-02-13 19:46:45 +00:00
|
|
|
return;
|
2009-06-25 11:52:33 +00:00
|
|
|
}
|
|
|
|
ifa_ref(&ia->ia_ifa);
|
2015-07-29 08:12:05 +00:00
|
|
|
IN_IFADDR_RUNLOCK(&in_ifa_tracker);
|
1997-02-13 19:46:45 +00:00
|
|
|
flags = RTF_UP;
|
|
|
|
ifp = ia->ia_ifa.ifa_ifp;
|
|
|
|
|
|
|
|
if ((ifp->if_flags & IFF_LOOPBACK)
|
|
|
|
|| (ifp->if_flags & IFF_POINTOPOINT))
|
|
|
|
flags |= RTF_HOST;
|
|
|
|
|
2011-05-20 19:12:20 +00:00
|
|
|
err = ifa_del_loopback_route((struct ifaddr *)ia, sa);
|
|
|
|
|
1997-02-13 19:46:45 +00:00
|
|
|
err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
|
|
|
|
if (err == 0)
|
|
|
|
ia->ia_flags |= IFA_ROUTE;
|
2011-05-20 19:12:20 +00:00
|
|
|
|
2009-09-15 19:18:34 +00:00
|
|
|
err = ifa_add_loopback_route((struct ifaddr *)ia, sa);
|
2011-05-20 19:12:20 +00:00
|
|
|
|
2009-06-25 11:52:33 +00:00
|
|
|
ifa_free(&ia->ia_ifa);
|
1997-02-13 19:46:45 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
1997-02-18 20:46:36 +00:00
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
rip_attach(struct socket *so, int proto, struct thread *td)
|
1997-02-18 20:46:36 +00:00
|
|
|
{
|
|
|
|
struct inpcb *inp;
|
2003-09-01 04:23:48 +00:00
|
|
|
int error;
|
1997-02-18 20:46:36 +00:00
|
|
|
|
|
|
|
inp = sotoinpcb(so);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
KASSERT(inp == NULL, ("rip_attach: inp != NULL"));
|
2007-06-12 00:12:01 +00:00
|
|
|
|
|
|
|
error = priv_check(td, PRIV_NETINET_RAW);
|
2006-11-06 13:42:10 +00:00
|
|
|
if (error)
|
2008-07-05 18:03:39 +00:00
|
|
|
return (error);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
if (proto >= IPPROTO_MAX || proto < 0)
|
2002-10-29 16:46:13 +00:00
|
|
|
return EPROTONOSUPPORT;
|
1999-12-22 19:13:38 +00:00
|
|
|
error = soreserve(so, rip_sendspace, rip_recvspace);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
if (error)
|
2008-07-05 18:03:39 +00:00
|
|
|
return (error);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WLOCK(&V_ripcbinfo);
|
|
|
|
error = in_pcballoc(so, &V_ripcbinfo);
|
2003-09-01 04:23:48 +00:00
|
|
|
if (error) {
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (error);
|
2003-09-01 04:23:48 +00:00
|
|
|
}
|
1997-02-18 20:46:36 +00:00
|
|
|
inp = (struct inpcb *)so->so_pcb;
|
1999-12-22 19:13:38 +00:00
|
|
|
inp->inp_vflag |= INP_IPV4;
|
1997-04-03 05:14:45 +00:00
|
|
|
inp->inp_ip_p = proto;
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
inp->inp_ip_ttl = V_ip_defttl;
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_inshash(inp);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WUNLOCK(inp);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (0);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2003-11-08 22:53:41 +00:00
|
|
|
static void
|
2006-07-21 17:11:15 +00:00
|
|
|
rip_detach(struct socket *so)
|
2003-11-08 22:53:41 +00:00
|
|
|
{
|
2006-07-21 17:11:15 +00:00
|
|
|
struct inpcb *inp;
|
2005-06-01 11:38:35 +00:00
|
|
|
|
2006-07-21 17:11:15 +00:00
|
|
|
inp = sotoinpcb(so);
|
|
|
|
KASSERT(inp != NULL, ("rip_detach: inp == NULL"));
|
|
|
|
KASSERT(inp->inp_faddr.s_addr == INADDR_ANY,
|
|
|
|
("rip_detach: not closed"));
|
2003-11-08 22:53:41 +00:00
|
|
|
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WLOCK(&V_ripcbinfo);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WLOCK(inp);
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_delhash(inp);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
if (so == V_ip_mrouter && ip_mrouter_done)
|
2003-11-08 22:53:41 +00:00
|
|
|
ip_mrouter_done();
|
|
|
|
if (ip_rsvp_force_done)
|
|
|
|
ip_rsvp_force_done(so);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
if (so == V_ip_rsvpd)
|
2003-11-08 22:53:41 +00:00
|
|
|
ip_rsvp_done();
|
|
|
|
in_pcbdetach(inp);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
in_pcbfree(inp);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
2003-11-08 22:53:41 +00:00
|
|
|
}
|
|
|
|
|
Chance protocol switch method pru_detach() so that it returns void
rather than an error. Detaches do not "fail", they other occur or
the protocol flags SS_PROTOREF to take ownership of the socket.
soclose() no longer looks at so_pcb to see if it's NULL, relying
entirely on the protocol to decide whether it's time to free the
socket or not using SS_PROTOREF. so_pcb is now entirely owned and
managed by the protocol code. Likewise, no longer test so_pcb in
other socket functions, such as soreceive(), which have no business
digging into protocol internals.
Protocol detach routines no longer try to free the socket on detach,
this is performed in the socket code if the protocol permits it.
In rts_detach(), no longer test for rp != NULL in detach, and
likewise in other protocols that don't permit a NULL so_pcb, reduce
the incidence of testing for it during detach.
netinet and netinet6 are not fully updated to this change, which
will be in an upcoming commit. In their current state they may leak
memory or panic.
MFC after: 3 months
2006-04-01 15:42:02 +00:00
|
|
|
static void
|
2006-07-21 17:11:15 +00:00
|
|
|
rip_dodisconnect(struct socket *so, struct inpcb *inp)
|
|
|
|
{
|
Decompose the current single inpcbinfo lock into two locks:
- The existing ipi_lock continues to protect the global inpcb list and
inpcb counter. This lock is now relegated to a small number of
allocation and free operations, and occasional operations that walk
all connections (including, awkwardly, certain UDP multicast receive
operations -- something to revisit).
- A new ipi_hash_lock protects the two inpcbinfo hash tables for
looking up connections and bound sockets, manipulated using new
INP_HASH_*() macros. This lock, combined with inpcb locks, protects
the 4-tuple address space.
Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required. As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.
A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb. Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed. In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup. New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:
INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb
Callers must pass exactly one of these flags (for the time being).
Some notes:
- All protocols are updated to work within the new regime; especially,
TCP, UDPv4, and UDPv6. pcbinfo ipi_lock acquisitions are largely
eliminated, and global hash lock hold times are dramatically reduced
compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
is no longer available -- hash lookup locks are now held only very
briefly during inpcb lookup, rather than for potentially extended
periods. However, the pcbinfo ipi_lock will still be acquired if a
connection state might change such that a connection is added or
removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
callers to acquire hash locks and perform one or more lookups atomically
with 4-tuple allocation: this is required only for TCPv6, as there is no
in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
locking, which relates to source address selection. This needs
attention, as it likely significantly reduces parallelism in this code
for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
is no longer sufficient. A second check once the inpcb lock is held
should do the trick, keeping the general case from requiring the inpcb
lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
undesirable, and probably another argument is required to take care of
this (or a char array name field in the pcbinfo?).
This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics. It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.
Reviewed by: bz
Sponsored by: Juniper Networks, Inc.
2011-05-30 09:43:55 +00:00
|
|
|
struct inpcbinfo *pcbinfo;
|
2008-07-28 06:57:28 +00:00
|
|
|
|
Decompose the current single inpcbinfo lock into two locks:
- The existing ipi_lock continues to protect the global inpcb list and
inpcb counter. This lock is now relegated to a small number of
allocation and free operations, and occasional operations that walk
all connections (including, awkwardly, certain UDP multicast receive
operations -- something to revisit).
- A new ipi_hash_lock protects the two inpcbinfo hash tables for
looking up connections and bound sockets, manipulated using new
INP_HASH_*() macros. This lock, combined with inpcb locks, protects
the 4-tuple address space.
Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required. As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.
A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb. Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed. In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup. New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:
INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb
Callers must pass exactly one of these flags (for the time being).
Some notes:
- All protocols are updated to work within the new regime; especially,
TCP, UDPv4, and UDPv6. pcbinfo ipi_lock acquisitions are largely
eliminated, and global hash lock hold times are dramatically reduced
compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
is no longer available -- hash lookup locks are now held only very
briefly during inpcb lookup, rather than for potentially extended
periods. However, the pcbinfo ipi_lock will still be acquired if a
connection state might change such that a connection is added or
removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
callers to acquire hash locks and perform one or more lookups atomically
with 4-tuple allocation: this is required only for TCPv6, as there is no
in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
locking, which relates to source address selection. This needs
attention, as it likely significantly reduces parallelism in this code
for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
is no longer sufficient. A second check once the inpcb lock is held
should do the trick, keeping the general case from requiring the inpcb
lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
undesirable, and probably another argument is required to take care of
this (or a char array name field in the pcbinfo?).
This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics. It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.
Reviewed by: bz
Sponsored by: Juniper Networks, Inc.
2011-05-30 09:43:55 +00:00
|
|
|
pcbinfo = inp->inp_pcbinfo;
|
|
|
|
INP_INFO_WLOCK(pcbinfo);
|
|
|
|
INP_WLOCK(inp);
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_delhash(inp);
|
2006-07-21 17:11:15 +00:00
|
|
|
inp->inp_faddr.s_addr = INADDR_ANY;
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_inshash(inp);
|
2006-07-21 17:11:15 +00:00
|
|
|
SOCK_LOCK(so);
|
|
|
|
so->so_state &= ~SS_ISCONNECTED;
|
|
|
|
SOCK_UNLOCK(so);
|
Decompose the current single inpcbinfo lock into two locks:
- The existing ipi_lock continues to protect the global inpcb list and
inpcb counter. This lock is now relegated to a small number of
allocation and free operations, and occasional operations that walk
all connections (including, awkwardly, certain UDP multicast receive
operations -- something to revisit).
- A new ipi_hash_lock protects the two inpcbinfo hash tables for
looking up connections and bound sockets, manipulated using new
INP_HASH_*() macros. This lock, combined with inpcb locks, protects
the 4-tuple address space.
Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required. As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.
A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb. Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed. In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup. New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:
INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb
Callers must pass exactly one of these flags (for the time being).
Some notes:
- All protocols are updated to work within the new regime; especially,
TCP, UDPv4, and UDPv6. pcbinfo ipi_lock acquisitions are largely
eliminated, and global hash lock hold times are dramatically reduced
compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
is no longer available -- hash lookup locks are now held only very
briefly during inpcb lookup, rather than for potentially extended
periods. However, the pcbinfo ipi_lock will still be acquired if a
connection state might change such that a connection is added or
removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
callers to acquire hash locks and perform one or more lookups atomically
with 4-tuple allocation: this is required only for TCPv6, as there is no
in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
locking, which relates to source address selection. This needs
attention, as it likely significantly reduces parallelism in this code
for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
is no longer sufficient. A second check once the inpcb lock is held
should do the trick, keeping the general case from requiring the inpcb
lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
undesirable, and probably another argument is required to take care of
this (or a char array name field in the pcbinfo?).
This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics. It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.
Reviewed by: bz
Sponsored by: Juniper Networks, Inc.
2011-05-30 09:43:55 +00:00
|
|
|
INP_WUNLOCK(inp);
|
|
|
|
INP_INFO_WUNLOCK(pcbinfo);
|
2006-07-21 17:11:15 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
rip_abort(struct socket *so)
|
1997-02-18 20:46:36 +00:00
|
|
|
{
|
|
|
|
struct inpcb *inp;
|
|
|
|
|
|
|
|
inp = sotoinpcb(so);
|
2006-07-21 17:11:15 +00:00
|
|
|
KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
|
|
|
|
|
|
|
|
rip_dodisconnect(so, inp);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2006-04-01 15:15:05 +00:00
|
|
|
static void
|
2006-07-21 17:11:15 +00:00
|
|
|
rip_close(struct socket *so)
|
1997-02-18 20:46:36 +00:00
|
|
|
{
|
2003-11-08 22:53:41 +00:00
|
|
|
struct inpcb *inp;
|
|
|
|
|
|
|
|
inp = sotoinpcb(so);
|
2006-07-21 17:11:15 +00:00
|
|
|
KASSERT(inp != NULL, ("rip_close: inp == NULL"));
|
|
|
|
|
|
|
|
rip_dodisconnect(so, inp);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1997-02-18 20:46:36 +00:00
|
|
|
static int
|
|
|
|
rip_disconnect(struct socket *so)
|
|
|
|
{
|
2006-05-15 09:28:57 +00:00
|
|
|
struct inpcb *inp;
|
|
|
|
|
2002-05-31 11:52:35 +00:00
|
|
|
if ((so->so_state & SS_ISCONNECTED) == 0)
|
2008-07-05 18:03:39 +00:00
|
|
|
return (ENOTCONN);
|
2006-05-15 09:28:57 +00:00
|
|
|
|
|
|
|
inp = sotoinpcb(so);
|
|
|
|
KASSERT(inp != NULL, ("rip_disconnect: inp == NULL"));
|
2008-07-05 18:03:39 +00:00
|
|
|
|
2006-07-21 17:11:15 +00:00
|
|
|
rip_dodisconnect(so, inp);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
return (0);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1997-02-18 20:46:36 +00:00
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
|
1997-02-18 20:46:36 +00:00
|
|
|
{
|
1997-08-16 19:16:27 +00:00
|
|
|
struct sockaddr_in *addr = (struct sockaddr_in *)nam;
|
2003-11-08 22:53:41 +00:00
|
|
|
struct inpcb *inp;
|
2009-02-05 14:06:09 +00:00
|
|
|
int error;
|
1997-02-18 20:46:36 +00:00
|
|
|
|
1997-08-16 19:16:27 +00:00
|
|
|
if (nam->sa_len != sizeof(*addr))
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EINVAL);
|
1997-02-18 20:46:36 +00:00
|
|
|
|
2009-02-05 14:06:09 +00:00
|
|
|
error = prison_check_ip4(td->td_ucred, &addr->sin_addr);
|
|
|
|
if (error != 0)
|
|
|
|
return (error);
|
2004-04-26 19:46:52 +00:00
|
|
|
|
2009-06-01 10:30:00 +00:00
|
|
|
inp = sotoinpcb(so);
|
|
|
|
KASSERT(inp != NULL, ("rip_bind: inp == NULL"));
|
|
|
|
|
2018-05-23 21:02:14 +00:00
|
|
|
if (CK_STAILQ_EMPTY(&V_ifnet) ||
|
2003-11-08 22:53:41 +00:00
|
|
|
(addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
|
2002-11-20 19:00:54 +00:00
|
|
|
(addr->sin_addr.s_addr &&
|
2009-06-01 10:30:00 +00:00
|
|
|
(inp->inp_flags & INP_BINDANY) == 0 &&
|
2009-06-22 10:59:34 +00:00
|
|
|
ifa_ifwithaddr_check((struct sockaddr *)addr) == 0))
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EADDRNOTAVAIL);
|
2003-11-08 22:53:41 +00:00
|
|
|
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WLOCK(&V_ripcbinfo);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WLOCK(inp);
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_delhash(inp);
|
1997-02-18 20:46:36 +00:00
|
|
|
inp->inp_laddr = addr->sin_addr;
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_inshash(inp);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WUNLOCK(inp);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (0);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1997-02-18 20:46:36 +00:00
|
|
|
static int
|
2001-09-12 08:38:13 +00:00
|
|
|
rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
|
1997-02-18 20:46:36 +00:00
|
|
|
{
|
1997-08-16 19:16:27 +00:00
|
|
|
struct sockaddr_in *addr = (struct sockaddr_in *)nam;
|
2003-11-08 22:53:41 +00:00
|
|
|
struct inpcb *inp;
|
1997-02-18 20:46:36 +00:00
|
|
|
|
1997-08-16 19:16:27 +00:00
|
|
|
if (nam->sa_len != sizeof(*addr))
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EINVAL);
|
2018-05-23 21:02:14 +00:00
|
|
|
if (CK_STAILQ_EMPTY(&V_ifnet))
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EADDRNOTAVAIL);
|
2003-11-08 22:53:41 +00:00
|
|
|
if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EAFNOSUPPORT);
|
2003-11-08 22:53:41 +00:00
|
|
|
|
|
|
|
inp = sotoinpcb(so);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
KASSERT(inp != NULL, ("rip_connect: inp == NULL"));
|
2008-07-05 18:03:39 +00:00
|
|
|
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WLOCK(&V_ripcbinfo);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WLOCK(inp);
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_delhash(inp);
|
1997-02-18 20:46:36 +00:00
|
|
|
inp->inp_faddr = addr->sin_addr;
|
2008-07-26 17:32:15 +00:00
|
|
|
rip_inshash(inp);
|
1997-02-18 20:46:36 +00:00
|
|
|
soisconnected(so);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WUNLOCK(inp);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (0);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1997-02-18 20:46:36 +00:00
|
|
|
static int
|
|
|
|
rip_shutdown(struct socket *so)
|
|
|
|
{
|
2003-11-08 22:53:41 +00:00
|
|
|
struct inpcb *inp;
|
|
|
|
|
|
|
|
inp = sotoinpcb(so);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
KASSERT(inp != NULL, ("rip_shutdown: inp == NULL"));
|
2008-07-05 18:03:39 +00:00
|
|
|
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WLOCK(inp);
|
1997-02-18 20:46:36 +00:00
|
|
|
socantsendmore(so);
|
2008-04-17 21:38:18 +00:00
|
|
|
INP_WUNLOCK(inp);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (0);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1994-05-24 10:09:53 +00:00
|
|
|
|
1997-02-18 20:46:36 +00:00
|
|
|
static int
|
1997-08-16 19:16:27 +00:00
|
|
|
rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
|
2007-05-10 15:58:48 +00:00
|
|
|
struct mbuf *control, struct thread *td)
|
1997-02-18 20:46:36 +00:00
|
|
|
{
|
2003-11-08 22:53:41 +00:00
|
|
|
struct inpcb *inp;
|
|
|
|
u_long dst;
|
1994-05-24 10:09:53 +00:00
|
|
|
|
2003-11-08 22:53:41 +00:00
|
|
|
inp = sotoinpcb(so);
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
KASSERT(inp != NULL, ("rip_send: inp == NULL"));
|
2008-07-05 18:03:39 +00:00
|
|
|
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
/*
|
|
|
|
* Note: 'dst' reads below are unlocked.
|
|
|
|
*/
|
1997-02-18 20:46:36 +00:00
|
|
|
if (so->so_state & SS_ISCONNECTED) {
|
|
|
|
if (nam) {
|
|
|
|
m_freem(m);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EISCONN);
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
Update in_pcb-derived basic socket types following changes to
pru_abort(), pru_detach(), and in_pcbdetach():
- Universally support and enforce the invariant that so_pcb is
never NULL, converting dozens of unnecessary NULL checks into
assertions, and eliminating dozens of unnecessary error handling
cases in protocol code.
- In some cases, eliminate unnecessary pcbinfo locking, as it is no
longer required to ensure so_pcb != NULL. For example, in protocol
shutdown methods, and in raw IP send.
- Abort and detach protocol switch methods no longer return failures,
nor attempt to free sockets, as the socket layer does this.
- Invoke in_pcbfree() after in_pcbdetach() in order to free the
detached in_pcb structure for a socket.
MFC after: 3 months
2006-04-01 16:20:54 +00:00
|
|
|
dst = inp->inp_faddr.s_addr; /* Unlocked read. */
|
1997-02-18 20:46:36 +00:00
|
|
|
} else {
|
|
|
|
if (nam == NULL) {
|
|
|
|
m_freem(m);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (ENOTCONN);
|
1997-02-18 20:46:36 +00:00
|
|
|
}
|
1997-08-16 19:16:27 +00:00
|
|
|
dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2008-07-05 18:03:39 +00:00
|
|
|
return (rip_output(m, so, dst));
|
1994-05-24 10:09:53 +00:00
|
|
|
}
|
2011-04-20 08:03:22 +00:00
|
|
|
#endif /* INET */
|
1997-02-18 20:46:36 +00:00
|
|
|
|
1998-05-15 20:11:40 +00:00
|
|
|
static int
|
2000-07-04 11:25:35 +00:00
|
|
|
rip_pcblist(SYSCTL_HANDLER_ARGS)
|
1998-05-15 20:11:40 +00:00
|
|
|
{
|
2003-09-01 04:23:48 +00:00
|
|
|
int error, i, n;
|
1998-05-15 20:11:40 +00:00
|
|
|
struct inpcb *inp, **inp_list;
|
|
|
|
inp_gen_t gencnt;
|
|
|
|
struct xinpgen xig;
|
2018-07-04 02:47:16 +00:00
|
|
|
struct epoch_tracker et;
|
1998-05-15 20:11:40 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The process of preparing the TCB list is too time-consuming and
|
|
|
|
* resource-intensive to repeat twice on every request.
|
|
|
|
*/
|
|
|
|
if (req->oldptr == 0) {
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
n = V_ripcbinfo.ipi_count;
|
2010-08-17 16:41:16 +00:00
|
|
|
n += imax(n / 8, 10);
|
|
|
|
req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (0);
|
1998-05-15 20:11:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (req->newptr != 0)
|
2008-07-05 18:03:39 +00:00
|
|
|
return (EPERM);
|
1998-05-15 20:11:40 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* OK, now we're committed to doing something.
|
|
|
|
*/
|
2018-07-04 02:47:16 +00:00
|
|
|
INP_INFO_WLOCK(&V_ripcbinfo);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
gencnt = V_ripcbinfo.ipi_gencnt;
|
|
|
|
n = V_ripcbinfo.ipi_count;
|
2018-07-04 02:47:16 +00:00
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
1998-05-15 20:11:40 +00:00
|
|
|
|
2018-11-22 20:49:41 +00:00
|
|
|
bzero(&xig, sizeof(xig));
|
1998-05-15 20:11:40 +00:00
|
|
|
xig.xig_len = sizeof xig;
|
|
|
|
xig.xig_count = n;
|
|
|
|
xig.xig_gen = gencnt;
|
|
|
|
xig.xig_sogen = so_gencnt;
|
|
|
|
error = SYSCTL_OUT(req, &xig, sizeof xig);
|
|
|
|
if (error)
|
2008-07-05 18:03:39 +00:00
|
|
|
return (error);
|
1998-05-15 20:11:40 +00:00
|
|
|
|
2018-07-02 05:19:44 +00:00
|
|
|
inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
|
2008-04-17 21:38:18 +00:00
|
|
|
|
2018-07-04 02:47:16 +00:00
|
|
|
INP_INFO_RLOCK_ET(&V_ripcbinfo, et);
|
2018-06-12 22:18:20 +00:00
|
|
|
for (inp = CK_LIST_FIRST(V_ripcbinfo.ipi_listhead), i = 0; inp && i < n;
|
|
|
|
inp = CK_LIST_NEXT(inp, inp_list)) {
|
2010-03-17 18:28:27 +00:00
|
|
|
INP_WLOCK(inp);
|
2003-09-17 21:13:16 +00:00
|
|
|
if (inp->inp_gencnt <= gencnt &&
|
2008-10-17 16:26:16 +00:00
|
|
|
cr_canseeinpcb(req->td->td_ucred, inp) == 0) {
|
2010-03-17 18:28:27 +00:00
|
|
|
in_pcbref(inp);
|
1998-05-15 20:11:40 +00:00
|
|
|
inp_list[i++] = inp;
|
2001-10-05 07:06:32 +00:00
|
|
|
}
|
2010-03-17 18:28:27 +00:00
|
|
|
INP_WUNLOCK(inp);
|
1998-05-15 20:11:40 +00:00
|
|
|
}
|
2018-07-04 02:47:16 +00:00
|
|
|
INP_INFO_RUNLOCK_ET(&V_ripcbinfo, et);
|
1998-05-15 20:11:40 +00:00
|
|
|
n = i;
|
|
|
|
|
|
|
|
error = 0;
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
inp = inp_list[i];
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RLOCK(inp);
|
1998-05-15 20:11:40 +00:00
|
|
|
if (inp->inp_gencnt <= gencnt) {
|
|
|
|
struct xinpcb xi;
|
2008-12-16 03:18:59 +00:00
|
|
|
|
Hide struct inpcb, struct tcpcb from the userland.
This is a painful change, but it is needed. On the one hand, we avoid
modifying them, and this slows down some ideas, on the other hand we still
eventually modify them and tools like netstat(1) never work on next version of
FreeBSD. We maintain a ton of spares in them, and we already got some ifdef
hell at the end of tcpcb.
Details:
- Hide struct inpcb, struct tcpcb under _KERNEL || _WANT_FOO.
- Make struct xinpcb, struct xtcpcb pure API structures, not including
kernel structures inpcb and tcpcb inside. Export into these structures
the fields from inpcb and tcpcb that are known to be used, and put there
a ton of spare space.
- Make kernel and userland utilities compilable after these changes.
- Bump __FreeBSD_version.
Reviewed by: rrs, gnn
Differential Revision: D10018
2017-03-21 06:39:49 +00:00
|
|
|
in_pcbtoxinpcb(inp, &xi);
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RUNLOCK(inp);
|
1998-05-15 20:11:40 +00:00
|
|
|
error = SYSCTL_OUT(req, &xi, sizeof xi);
|
2006-07-18 22:34:27 +00:00
|
|
|
} else
|
2008-04-21 12:06:41 +00:00
|
|
|
INP_RUNLOCK(inp);
|
1998-05-15 20:11:40 +00:00
|
|
|
}
|
2018-07-02 05:19:44 +00:00
|
|
|
INP_INFO_WLOCK(&V_ripcbinfo);
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
inp = inp_list[i];
|
|
|
|
INP_RLOCK(inp);
|
|
|
|
if (!in_pcbrele_rlocked(inp))
|
|
|
|
INP_RUNLOCK(inp);
|
|
|
|
}
|
|
|
|
INP_INFO_WUNLOCK(&V_ripcbinfo);
|
2010-03-17 18:28:27 +00:00
|
|
|
|
1998-05-15 20:11:40 +00:00
|
|
|
if (!error) {
|
2018-07-04 02:47:16 +00:00
|
|
|
struct epoch_tracker et;
|
1998-05-15 20:11:40 +00:00
|
|
|
/*
|
2008-07-05 18:03:39 +00:00
|
|
|
* Give the user an updated idea of our state. If the
|
|
|
|
* generation differs from what we told her before, she knows
|
|
|
|
* that something happened while we were processing this
|
|
|
|
* request, and it might be necessary to retry.
|
1998-05-15 20:11:40 +00:00
|
|
|
*/
|
2018-07-04 02:47:16 +00:00
|
|
|
INP_INFO_RLOCK_ET(&V_ripcbinfo, et);
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
xig.xig_gen = V_ripcbinfo.ipi_gencnt;
|
1998-05-15 20:11:40 +00:00
|
|
|
xig.xig_sogen = so_gencnt;
|
Commit step 1 of the vimage project, (network stack)
virtualization work done by Marko Zec (zec@).
This is the first in a series of commits over the course
of the next few weeks.
Mark all uses of global variables to be virtualized
with a V_ prefix.
Use macros to map them back to their global names for
now, so this is a NOP change only.
We hope to have caught at least 85-90% of what is needed
so we do not invalidate a lot of outstanding patches again.
Obtained from: //depot/projects/vimage-commit2/...
Reviewed by: brooks, des, ed, mav, julian,
jamie, kris, rwatson, zec, ...
(various people I forgot, different versions)
md5 (with a bit of help)
Sponsored by: NLnet Foundation, The FreeBSD Foundation
X-MFC after: never
V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
|
|
|
xig.xig_count = V_ripcbinfo.ipi_count;
|
2018-07-04 02:47:16 +00:00
|
|
|
INP_INFO_RUNLOCK_ET(&V_ripcbinfo, et);
|
1998-05-15 20:11:40 +00:00
|
|
|
error = SYSCTL_OUT(req, &xig, sizeof xig);
|
|
|
|
}
|
2018-07-02 05:19:44 +00:00
|
|
|
free(inp_list, M_TEMP);
|
2008-07-05 18:03:39 +00:00
|
|
|
return (error);
|
1998-05-15 20:11:40 +00:00
|
|
|
}
|
|
|
|
|
2011-01-18 21:14:13 +00:00
|
|
|
SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist,
|
|
|
|
CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0,
|
2008-07-05 18:03:39 +00:00
|
|
|
rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
|
1998-05-15 20:11:40 +00:00
|
|
|
|
2011-04-20 08:03:22 +00:00
|
|
|
#ifdef INET
|
1997-02-18 20:46:36 +00:00
|
|
|
struct pr_usrreqs rip_usrreqs = {
|
2004-11-08 14:44:54 +00:00
|
|
|
.pru_abort = rip_abort,
|
|
|
|
.pru_attach = rip_attach,
|
|
|
|
.pru_bind = rip_bind,
|
|
|
|
.pru_connect = rip_connect,
|
|
|
|
.pru_control = in_control,
|
|
|
|
.pru_detach = rip_detach,
|
|
|
|
.pru_disconnect = rip_disconnect,
|
2007-05-11 10:20:51 +00:00
|
|
|
.pru_peeraddr = in_getpeeraddr,
|
2004-11-08 14:44:54 +00:00
|
|
|
.pru_send = rip_send,
|
|
|
|
.pru_shutdown = rip_shutdown,
|
2007-05-11 10:20:51 +00:00
|
|
|
.pru_sockaddr = in_getsockaddr,
|
2006-07-21 17:11:15 +00:00
|
|
|
.pru_sosetlabel = in_pcbsosetlabel,
|
|
|
|
.pru_close = rip_close,
|
1997-02-18 20:46:36 +00:00
|
|
|
};
|
2011-04-20 08:03:22 +00:00
|
|
|
#endif /* INET */
|