2003-06-23 16:55:01 +00:00
|
|
|
/*-
|
2003-06-27 05:13:52 +00:00
|
|
|
* Copyright (c) 2001 Atsushi Onoe
|
2009-01-08 17:12:47 +00:00
|
|
|
* Copyright (c) 2002-2009 Sam Leffler, Errno Consulting
|
2003-06-23 16:55:01 +00:00
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* 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
|
2003-06-27 05:13:52 +00:00
|
|
|
* 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.
|
2003-06-23 16:55:01 +00:00
|
|
|
*
|
2003-06-27 05:13:52 +00:00
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
|
2003-06-23 16:55:01 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
|
|
|
#include "opt_inet.h"
|
2009-06-07 22:06:15 +00:00
|
|
|
#include "opt_inet6.h"
|
2008-04-20 20:35:46 +00:00
|
|
|
#include "opt_wlan.h"
|
2003-06-23 16:55:01 +00:00
|
|
|
|
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
#include <sys/mbuf.h>
|
|
|
|
#include <sys/kernel.h>
|
|
|
|
#include <sys/endian.h>
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
#include <sys/socket.h>
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
#include <net/bpf.h>
|
2003-06-23 16:55:01 +00:00
|
|
|
#include <net/ethernet.h>
|
2004-12-08 17:26:47 +00:00
|
|
|
#include <net/if.h>
|
2013-10-26 17:58:36 +00:00
|
|
|
#include <net/if_var.h>
|
2003-06-23 16:55:01 +00:00
|
|
|
#include <net/if_llc.h>
|
2004-12-08 17:26:47 +00:00
|
|
|
#include <net/if_media.h>
|
|
|
|
#include <net/if_vlan_var.h>
|
2003-06-23 16:55:01 +00:00
|
|
|
|
|
|
|
#include <net80211/ieee80211_var.h>
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
#include <net80211/ieee80211_regdomain.h>
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
|
|
|
#include <net80211/ieee80211_superg.h>
|
|
|
|
#endif
|
2009-01-08 17:12:47 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_TDMA
|
|
|
|
#include <net80211/ieee80211_tdma.h>
|
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
#include <net80211/ieee80211_wds.h>
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#include <net80211/ieee80211_mesh.h>
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2011-04-25 16:37:47 +00:00
|
|
|
#if defined(INET) || defined(INET6)
|
2003-06-23 16:55:01 +00:00
|
|
|
#include <netinet/in.h>
|
2011-04-25 16:37:47 +00:00
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef INET
|
2003-06-23 16:55:01 +00:00
|
|
|
#include <netinet/if_ether.h>
|
2004-12-08 17:26:47 +00:00
|
|
|
#include <netinet/in_systm.h>
|
|
|
|
#include <netinet/ip.h>
|
|
|
|
#endif
|
2009-06-07 22:06:15 +00:00
|
|
|
#ifdef INET6
|
|
|
|
#include <netinet/ip6.h>
|
|
|
|
#endif
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2009-06-05 13:55:33 +00:00
|
|
|
#include <security/mac/mac_framework.h>
|
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
#define ETHER_HEADER_COPY(dst, src) \
|
|
|
|
memcpy(dst, src, sizeof(struct ether_header))
|
|
|
|
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
/* unalligned little endian access */
|
|
|
|
#define LE_WRITE_2(p, v) do { \
|
|
|
|
((uint8_t *)(p))[0] = (v) & 0xff; \
|
|
|
|
((uint8_t *)(p))[1] = ((v) >> 8) & 0xff; \
|
|
|
|
} while (0)
|
|
|
|
#define LE_WRITE_4(p, v) do { \
|
|
|
|
((uint8_t *)(p))[0] = (v) & 0xff; \
|
|
|
|
((uint8_t *)(p))[1] = ((v) >> 8) & 0xff; \
|
|
|
|
((uint8_t *)(p))[2] = ((v) >> 16) & 0xff; \
|
|
|
|
((uint8_t *)(p))[3] = ((v) >> 24) & 0xff; \
|
|
|
|
} while (0)
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
static int ieee80211_fragment(struct ieee80211vap *, struct mbuf *,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
u_int hdrsize, u_int ciphdrsize, u_int mtu);
|
|
|
|
static void ieee80211_tx_mgt_cb(struct ieee80211_node *, void *, int);
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
#ifdef IEEE80211_DEBUG
|
|
|
|
/*
|
|
|
|
* Decide if an outbound management frame should be
|
|
|
|
* printed when debugging is enabled. This filters some
|
|
|
|
* of the less interesting frames that come frequently
|
|
|
|
* (e.g. beacons).
|
|
|
|
*/
|
|
|
|
static __inline int
|
2008-04-20 20:35:46 +00:00
|
|
|
doprint(struct ieee80211vap *vap, int subtype)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
|
|
|
switch (subtype) {
|
|
|
|
case IEEE80211_FC0_SUBTYPE_PROBE_RESP:
|
2008-04-20 20:35:46 +00:00
|
|
|
return (vap->iv_opmode == IEEE80211_M_IBSS);
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
return 1;
|
|
|
|
}
|
2003-06-23 16:55:01 +00:00
|
|
|
#endif
|
|
|
|
|
2013-07-28 04:53:00 +00:00
|
|
|
/*
|
|
|
|
* Transmit a frame to the given destination on the given VAP.
|
|
|
|
*
|
|
|
|
* It's up to the caller to figure out the details of who this
|
|
|
|
* is going to and resolving the node.
|
|
|
|
*
|
|
|
|
* This routine takes care of queuing it for power save,
|
|
|
|
* A-MPDU state stuff, fast-frames state stuff, encapsulation
|
|
|
|
* if required, then passing it up to the driver layer.
|
|
|
|
*
|
|
|
|
* This routine (for now) consumes the mbuf and frees the node
|
|
|
|
* reference; it ideally will return a TX status which reflects
|
|
|
|
* whether the mbuf was consumed or not, so the caller can
|
|
|
|
* free the mbuf (if appropriate) and the node reference (again,
|
|
|
|
* if appropriate.)
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
ieee80211_vap_pkt_send_dest(struct ieee80211vap *vap, struct mbuf *m,
|
|
|
|
struct ieee80211_node *ni)
|
|
|
|
{
|
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
|
|
|
struct ifnet *ifp = vap->iv_ifp;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
if ((ni->ni_flags & IEEE80211_NODE_PWR_MGT) &&
|
|
|
|
(m->m_flags & M_PWR_SAV) == 0) {
|
|
|
|
/*
|
|
|
|
* Station in power save mode; pass the frame
|
|
|
|
* to the 802.11 layer and continue. We'll get
|
|
|
|
* the frame back when the time is right.
|
|
|
|
* XXX lose WDS vap linkage?
|
|
|
|
*/
|
|
|
|
(void) ieee80211_pwrsave(ni, m);
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
/* calculate priority so drivers can find the tx queue */
|
|
|
|
if (ieee80211_classify(ni, m)) {
|
|
|
|
IEEE80211_DISCARD_MAC(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
ni->ni_macaddr, NULL,
|
|
|
|
"%s", "classification failure");
|
|
|
|
vap->iv_stats.is_tx_classify++;
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
m_freem(m);
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Stash the node pointer. Note that we do this after
|
|
|
|
* any call to ieee80211_dwds_mcast because that code
|
|
|
|
* uses any existing value for rcvif to identify the
|
|
|
|
* interface it (might have been) received on.
|
|
|
|
*/
|
|
|
|
m->m_pkthdr.rcvif = (void *)ni;
|
|
|
|
|
|
|
|
BPF_MTAP(ifp, m); /* 802.3 tx */
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if A-MPDU tx aggregation is setup or if we
|
|
|
|
* should try to enable it. The sta must be associated
|
|
|
|
* with HT and A-MPDU enabled for use. When the policy
|
|
|
|
* routine decides we should enable A-MPDU we issue an
|
|
|
|
* ADDBA request and wait for a reply. The frame being
|
|
|
|
* encapsulated will go out w/o using A-MPDU, or possibly
|
|
|
|
* it might be collected by the driver and held/retransmit.
|
|
|
|
* The default ic_ampdu_enable routine handles staggering
|
|
|
|
* ADDBA requests in case the receiver NAK's us or we are
|
|
|
|
* otherwise unable to establish a BA stream.
|
|
|
|
*/
|
|
|
|
if ((ni->ni_flags & IEEE80211_NODE_AMPDU_TX) &&
|
|
|
|
(vap->iv_flags_ht & IEEE80211_FHT_AMPDU_TX) &&
|
|
|
|
(m->m_flags & M_EAPOL) == 0) {
|
|
|
|
int tid = WME_AC_TO_TID(M_WME_GETAC(m));
|
|
|
|
struct ieee80211_tx_ampdu *tap = &ni->ni_tx_ampdu[tid];
|
|
|
|
|
|
|
|
ieee80211_txampdu_count_packet(tap);
|
|
|
|
if (IEEE80211_AMPDU_RUNNING(tap)) {
|
|
|
|
/*
|
|
|
|
* Operational, mark frame for aggregation.
|
|
|
|
*
|
|
|
|
* XXX do tx aggregation here
|
|
|
|
*/
|
|
|
|
m->m_flags |= M_AMPDU_MPDU;
|
|
|
|
} else if (!IEEE80211_AMPDU_REQUESTED(tap) &&
|
|
|
|
ic->ic_ampdu_enable(ni, tap)) {
|
|
|
|
/*
|
|
|
|
* Not negotiated yet, request service.
|
|
|
|
*/
|
|
|
|
ieee80211_ampdu_request(ni, tap);
|
|
|
|
/* XXX hold frame for reply? */
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
|
|
|
else if (IEEE80211_ATH_CAP(vap, ni, IEEE80211_NODE_FF)) {
|
|
|
|
m = ieee80211_ff_check(ni, m);
|
|
|
|
if (m == NULL) {
|
|
|
|
/* NB: any ni ref held on stageq */
|
2013-08-18 20:40:13 +00:00
|
|
|
return (0);
|
2013-07-28 04:53:00 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif /* IEEE80211_SUPPORT_SUPERG */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Grab the TX lock - serialise the TX process from this
|
|
|
|
* point (where TX state is being checked/modified)
|
|
|
|
* through to driver queue.
|
|
|
|
*/
|
|
|
|
IEEE80211_TX_LOCK(ic);
|
|
|
|
|
|
|
|
if (__predict_true((vap->iv_caps & IEEE80211_C_8023ENCAP) == 0)) {
|
|
|
|
/*
|
|
|
|
* Encapsulate the packet in prep for transmission.
|
|
|
|
*/
|
|
|
|
m = ieee80211_encap(vap, ni, m);
|
|
|
|
if (m == NULL) {
|
|
|
|
/* NB: stat+msg handled in ieee80211_encap */
|
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
}
|
2013-08-08 05:09:35 +00:00
|
|
|
error = ieee80211_parent_xmitpkt(ic, m);
|
2013-07-28 04:53:00 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Unlock at this point - no need to hold it across
|
|
|
|
* ieee80211_free_node() (ie, the comlock)
|
|
|
|
*/
|
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
if (error != 0) {
|
|
|
|
/* NB: IFQ_HANDOFF reclaims mbuf */
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
} else {
|
|
|
|
ifp->if_opackets++;
|
|
|
|
}
|
|
|
|
ic->ic_lastdata = ticks;
|
|
|
|
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
/*
|
|
|
|
* Send the given mbuf through the given vap.
|
|
|
|
*
|
|
|
|
* This consumes the mbuf regardless of whether the transmit
|
|
|
|
* was successful or not.
|
|
|
|
*
|
|
|
|
* This does none of the initial checks that ieee80211_start()
|
|
|
|
* does (eg CAC timeout, interface wakeup) - the caller must
|
|
|
|
* do this first.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
ieee80211_start_pkt(struct ieee80211vap *vap, struct mbuf *m)
|
|
|
|
{
|
|
|
|
#define IS_DWDS(vap) \
|
|
|
|
(vap->iv_opmode == IEEE80211_M_WDS && \
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_WDSLEGACY) == 0)
|
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
|
|
|
struct ifnet *ifp = vap->iv_ifp;
|
|
|
|
struct ieee80211_node *ni;
|
|
|
|
struct ether_header *eh;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Cancel any background scan.
|
|
|
|
*/
|
|
|
|
if (ic->ic_flags & IEEE80211_F_SCAN)
|
|
|
|
ieee80211_cancel_anyscan(vap);
|
|
|
|
/*
|
|
|
|
* Find the node for the destination so we can do
|
|
|
|
* things like power save and fast frames aggregation.
|
|
|
|
*
|
|
|
|
* NB: past this point various code assumes the first
|
|
|
|
* mbuf has the 802.3 header present (and contiguous).
|
|
|
|
*/
|
|
|
|
ni = NULL;
|
|
|
|
if (m->m_len < sizeof(struct ether_header) &&
|
|
|
|
(m = m_pullup(m, sizeof(struct ether_header))) == NULL) {
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
"discard frame, %s\n", "m_pullup failed");
|
|
|
|
vap->iv_stats.is_tx_nobuf++; /* XXX */
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
eh = mtod(m, struct ether_header *);
|
|
|
|
if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
|
|
|
|
if (IS_DWDS(vap)) {
|
|
|
|
/*
|
|
|
|
* Only unicast frames from the above go out
|
|
|
|
* DWDS vaps; multicast frames are handled by
|
|
|
|
* dispatching the frame as it comes through
|
|
|
|
* the AP vap (see below).
|
|
|
|
*/
|
|
|
|
IEEE80211_DISCARD_MAC(vap, IEEE80211_MSG_WDS,
|
|
|
|
eh->ether_dhost, "mcast", "%s", "on DWDS");
|
|
|
|
vap->iv_stats.is_dwds_mcast++;
|
|
|
|
m_freem(m);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_HOSTAP) {
|
|
|
|
/*
|
|
|
|
* Spam DWDS vap's w/ multicast traffic.
|
|
|
|
*/
|
|
|
|
/* XXX only if dwds in use? */
|
|
|
|
ieee80211_dwds_mcast(vap, m);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode != IEEE80211_M_MBSS) {
|
|
|
|
#endif
|
|
|
|
ni = ieee80211_find_txnode(vap, eh->ether_dhost);
|
|
|
|
if (ni == NULL) {
|
|
|
|
/* NB: ieee80211_find_txnode does stat+msg */
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
m_freem(m);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
if (ni->ni_associd == 0 &&
|
|
|
|
(ni->ni_flags & IEEE80211_NODE_ASSOCID)) {
|
|
|
|
IEEE80211_DISCARD_MAC(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
eh->ether_dhost, NULL,
|
|
|
|
"sta not associated (type 0x%04x)",
|
|
|
|
htons(eh->ether_type));
|
|
|
|
vap->iv_stats.is_tx_notassoc++;
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
m_freem(m);
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
} else {
|
|
|
|
if (!IEEE80211_ADDR_EQ(eh->ether_shost, vap->iv_myaddr)) {
|
|
|
|
/*
|
|
|
|
* Proxy station only if configured.
|
|
|
|
*/
|
|
|
|
if (!ieee80211_mesh_isproxyena(vap)) {
|
|
|
|
IEEE80211_DISCARD_MAC(vap,
|
|
|
|
IEEE80211_MSG_OUTPUT |
|
|
|
|
IEEE80211_MSG_MESH,
|
|
|
|
eh->ether_dhost, NULL,
|
|
|
|
"%s", "proxy not enabled");
|
|
|
|
vap->iv_stats.is_mesh_notproxy++;
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
m_freem(m);
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
"forward frame from DS SA(%6D), DA(%6D)\n",
|
|
|
|
eh->ether_shost, ":",
|
|
|
|
eh->ether_dhost, ":");
|
|
|
|
ieee80211_mesh_proxy_check(vap, eh->ether_shost);
|
|
|
|
}
|
|
|
|
ni = ieee80211_mesh_discover(vap, eh->ether_dhost, m);
|
|
|
|
if (ni == NULL) {
|
|
|
|
/*
|
|
|
|
* NB: ieee80211_mesh_discover holds/disposes
|
|
|
|
* frame (e.g. queueing on path discovery).
|
|
|
|
*/
|
|
|
|
ifp->if_oerrors++;
|
|
|
|
/* XXX better status? */
|
|
|
|
return (ENOBUFS);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/*
|
2013-07-28 04:53:00 +00:00
|
|
|
* We've resolved the sender, so attempt to transmit it.
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
*/
|
2013-07-28 04:53:00 +00:00
|
|
|
if (ieee80211_vap_pkt_send_dest(vap, m, ni) != 0)
|
|
|
|
return (ENOBUFS);
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
return (0);
|
|
|
|
#undef IS_DWDS
|
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Start method for vap's. All packets from the stack come
|
|
|
|
* through here. We handle common processing of the packets
|
|
|
|
* before dispatching them to the underlying device.
|
2013-08-10 21:46:58 +00:00
|
|
|
*
|
|
|
|
* if_transmit() requires that the mbuf be consumed by this call
|
|
|
|
* regardless of the return condition.
|
2008-04-20 20:35:46 +00:00
|
|
|
*/
|
2013-08-08 05:09:35 +00:00
|
|
|
int
|
|
|
|
ieee80211_vap_transmit(struct ifnet *ifp, struct mbuf *m)
|
2008-04-20 20:35:46 +00:00
|
|
|
{
|
|
|
|
struct ieee80211vap *vap = ifp->if_softc;
|
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
|
|
|
struct ifnet *parent = ic->ic_ifp;
|
|
|
|
|
|
|
|
/* NB: parent must be up and running */
|
|
|
|
if (!IFNET_IS_UP_RUNNING(parent)) {
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
"%s: ignore queue, parent %s not up+running\n",
|
|
|
|
__func__, parent->if_xname);
|
|
|
|
/* XXX stat */
|
2013-08-10 21:46:58 +00:00
|
|
|
m_freem(m);
|
2013-08-08 05:09:35 +00:00
|
|
|
return (EINVAL);
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
|
|
|
if (vap->iv_state == IEEE80211_S_SLEEP) {
|
|
|
|
/*
|
|
|
|
* In power save, wakeup device for transmit.
|
|
|
|
*/
|
|
|
|
ieee80211_new_state(vap, IEEE80211_S_RUN, 0);
|
2013-08-10 21:46:58 +00:00
|
|
|
m_freem(m);
|
2013-08-08 05:09:35 +00:00
|
|
|
return (0);
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
|
|
|
/*
|
|
|
|
* No data frames go out unless we're running.
|
|
|
|
* Note in particular this covers CAC and CSA
|
|
|
|
* states (though maybe we should check muting
|
|
|
|
* for CSA).
|
|
|
|
*/
|
|
|
|
if (vap->iv_state != IEEE80211_S_RUN) {
|
|
|
|
IEEE80211_LOCK(ic);
|
|
|
|
/* re-check under the com lock to avoid races */
|
|
|
|
if (vap->iv_state != IEEE80211_S_RUN) {
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
"%s: ignore queue, in %s state\n",
|
|
|
|
__func__, ieee80211_state_name[vap->iv_state]);
|
|
|
|
vap->iv_stats.is_tx_badstate++;
|
|
|
|
IEEE80211_UNLOCK(ic);
|
2012-02-24 05:40:36 +00:00
|
|
|
ifp->if_drv_flags |= IFF_DRV_OACTIVE;
|
2013-08-10 21:46:58 +00:00
|
|
|
m_freem(m);
|
2013-08-08 05:09:35 +00:00
|
|
|
return (EINVAL);
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
|
|
|
IEEE80211_UNLOCK(ic);
|
|
|
|
}
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
|
2013-08-08 05:09:35 +00:00
|
|
|
/*
|
|
|
|
* Sanitize mbuf flags for net80211 use. We cannot
|
|
|
|
* clear M_PWR_SAV or M_MORE_DATA because these may
|
|
|
|
* be set for frames that are re-submitted from the
|
|
|
|
* power save queue.
|
|
|
|
*
|
|
|
|
* NB: This must be done before ieee80211_classify as
|
|
|
|
* it marks EAPOL in frames with M_EAPOL.
|
|
|
|
*/
|
|
|
|
m->m_flags &= ~(M_80211_TX - M_PWR_SAV - M_MORE_DATA);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Bump to the packet transmission path.
|
|
|
|
* The mbuf will be consumed here.
|
|
|
|
*/
|
|
|
|
return (ieee80211_start_pkt(vap, m));
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
ieee80211_vap_qflush(struct ifnet *ifp)
|
|
|
|
{
|
|
|
|
|
|
|
|
/* Empty for now */
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 802.11 raw output routine.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
ieee80211_raw_output(struct ieee80211vap *vap, struct ieee80211_node *ni,
|
|
|
|
struct mbuf *m, const struct ieee80211_bpf_params *params)
|
|
|
|
{
|
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
|
|
|
|
|
|
|
return (ic->ic_raw_xmit(ni, m, params));
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 802.11 output routine. This is (currently) used only to
|
|
|
|
* connect bpf write calls to the 802.11 layer for injecting
|
2009-04-26 21:13:18 +00:00
|
|
|
* raw 802.11 frames.
|
2008-04-20 20:35:46 +00:00
|
|
|
*/
|
2013-08-07 22:01:43 +00:00
|
|
|
#if __FreeBSD_version >= 1000031
|
2008-04-20 20:35:46 +00:00
|
|
|
int
|
|
|
|
ieee80211_output(struct ifnet *ifp, struct mbuf *m,
|
2013-04-26 12:50:32 +00:00
|
|
|
const struct sockaddr *dst, struct route *ro)
|
2013-08-07 22:01:43 +00:00
|
|
|
#else
|
|
|
|
int
|
|
|
|
ieee80211_output(struct ifnet *ifp, struct mbuf *m,
|
|
|
|
struct sockaddr *dst, struct route *ro)
|
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
{
|
|
|
|
#define senderr(e) do { error = (e); goto bad;} while (0)
|
|
|
|
struct ieee80211_node *ni = NULL;
|
2009-07-24 15:37:02 +00:00
|
|
|
struct ieee80211vap *vap;
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211_frame *wh;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
struct ieee80211com *ic = NULL;
|
2008-04-20 20:35:46 +00:00
|
|
|
int error;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
int ret;
|
2008-04-20 20:35:46 +00:00
|
|
|
|
2009-07-24 15:37:02 +00:00
|
|
|
if (ifp->if_drv_flags & IFF_DRV_OACTIVE) {
|
|
|
|
/*
|
|
|
|
* Short-circuit requests if the vap is marked OACTIVE
|
|
|
|
* as this can happen because a packet came down through
|
|
|
|
* ieee80211_start before the vap entered RUN state in
|
|
|
|
* which case it's ok to just drop the frame. This
|
|
|
|
* should not be necessary but callers of if_output don't
|
|
|
|
* check OACTIVE.
|
|
|
|
*/
|
|
|
|
senderr(ENETDOWN);
|
|
|
|
}
|
|
|
|
vap = ifp->if_softc;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
ic = vap->iv_ic;
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Hand to the 802.3 code if not tagged as
|
|
|
|
* a raw 802.11 frame.
|
|
|
|
*/
|
|
|
|
if (dst->sa_family != AF_IEEE80211)
|
2009-04-16 20:30:28 +00:00
|
|
|
return vap->iv_output(ifp, m, dst, ro);
|
2008-04-20 20:35:46 +00:00
|
|
|
#ifdef MAC
|
2009-06-05 13:55:33 +00:00
|
|
|
error = mac_ifnet_check_transmit(ifp, m);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (error)
|
|
|
|
senderr(error);
|
|
|
|
#endif
|
|
|
|
if (ifp->if_flags & IFF_MONITOR)
|
|
|
|
senderr(ENETDOWN);
|
|
|
|
if (!IFNET_IS_UP_RUNNING(ifp))
|
|
|
|
senderr(ENETDOWN);
|
|
|
|
if (vap->iv_state == IEEE80211_S_CAC) {
|
|
|
|
IEEE80211_DPRINTF(vap,
|
|
|
|
IEEE80211_MSG_OUTPUT | IEEE80211_MSG_DOTH,
|
|
|
|
"block %s frame in CAC state\n", "raw data");
|
|
|
|
vap->iv_stats.is_tx_badstate++;
|
|
|
|
senderr(EIO); /* XXX */
|
2011-03-13 12:56:46 +00:00
|
|
|
} else if (vap->iv_state == IEEE80211_S_SCAN)
|
|
|
|
senderr(EIO);
|
2008-04-20 20:35:46 +00:00
|
|
|
/* XXX bypass bridge, pfil, carp, etc. */
|
|
|
|
|
|
|
|
if (m->m_pkthdr.len < sizeof(struct ieee80211_frame_ack))
|
|
|
|
senderr(EIO); /* XXX */
|
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
|
|
|
if ((wh->i_fc[0] & IEEE80211_FC0_VERSION_MASK) !=
|
|
|
|
IEEE80211_FC0_VERSION_0)
|
|
|
|
senderr(EIO); /* XXX */
|
|
|
|
|
|
|
|
/* locate destination node */
|
|
|
|
switch (wh->i_fc[1] & IEEE80211_FC1_DIR_MASK) {
|
|
|
|
case IEEE80211_FC1_DIR_NODS:
|
|
|
|
case IEEE80211_FC1_DIR_FROMDS:
|
|
|
|
ni = ieee80211_find_txnode(vap, wh->i_addr1);
|
|
|
|
break;
|
|
|
|
case IEEE80211_FC1_DIR_TODS:
|
|
|
|
case IEEE80211_FC1_DIR_DSTODS:
|
|
|
|
if (m->m_pkthdr.len < sizeof(struct ieee80211_frame))
|
|
|
|
senderr(EIO); /* XXX */
|
|
|
|
ni = ieee80211_find_txnode(vap, wh->i_addr3);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
senderr(EIO); /* XXX */
|
|
|
|
}
|
|
|
|
if (ni == NULL) {
|
|
|
|
/*
|
|
|
|
* Permit packets w/ bpf params through regardless
|
|
|
|
* (see below about sa_len).
|
|
|
|
*/
|
|
|
|
if (dst->sa_len == 0)
|
|
|
|
senderr(EHOSTUNREACH);
|
|
|
|
ni = ieee80211_ref_node(vap->iv_bss);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Sanitize mbuf for net80211 flags leaked from above.
|
|
|
|
*
|
|
|
|
* NB: This must be done before ieee80211_classify as
|
|
|
|
* it marks EAPOL in frames with M_EAPOL.
|
|
|
|
*/
|
|
|
|
m->m_flags &= ~M_80211_TX;
|
|
|
|
|
|
|
|
/* calculate priority so drivers can find the tx queue */
|
|
|
|
/* XXX assumes an 802.3 frame */
|
|
|
|
if (ieee80211_classify(ni, m))
|
|
|
|
senderr(EIO); /* XXX */
|
|
|
|
|
2009-07-24 15:22:12 +00:00
|
|
|
ifp->if_opackets++;
|
2009-04-26 21:11:12 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_data);
|
|
|
|
if (IEEE80211_IS_MULTICAST(wh->i_addr1)) {
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_mcast);
|
|
|
|
m->m_flags |= M_MCAST;
|
|
|
|
} else
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_ucast);
|
|
|
|
/* NB: ieee80211_encap does not include 802.11 header */
|
|
|
|
IEEE80211_NODE_STAT_ADD(ni, tx_bytes, m->m_pkthdr.len);
|
2008-04-20 20:35:46 +00:00
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
IEEE80211_TX_LOCK(ic);
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* NB: DLT_IEEE802_11_RADIO identifies the parameters are
|
|
|
|
* present by setting the sa_len field of the sockaddr (yes,
|
|
|
|
* this is a hack).
|
|
|
|
* NB: we assume sa_data is suitably aligned to cast.
|
|
|
|
*/
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
ret = ieee80211_raw_output(vap, ni, m,
|
2008-04-20 20:35:46 +00:00
|
|
|
(const struct ieee80211_bpf_params *)(dst->sa_len ?
|
|
|
|
dst->sa_data : NULL));
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
return (ret);
|
2008-04-20 20:35:46 +00:00
|
|
|
bad:
|
|
|
|
if (m != NULL)
|
|
|
|
m_freem(m);
|
|
|
|
if (ni != NULL)
|
|
|
|
ieee80211_free_node(ni);
|
2009-07-24 15:22:12 +00:00
|
|
|
ifp->if_oerrors++;
|
2008-04-20 20:35:46 +00:00
|
|
|
return error;
|
|
|
|
#undef senderr
|
|
|
|
}
|
|
|
|
|
2005-07-22 20:48:11 +00:00
|
|
|
/*
|
|
|
|
* Set the direction field and address fields of an outgoing
|
2008-10-26 00:52:26 +00:00
|
|
|
* frame. Note this should be called early on in constructing
|
|
|
|
* a frame as it sets i_fc[1]; other bits can then be or'd in.
|
2005-07-22 20:48:11 +00:00
|
|
|
*/
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
void
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_send_setup(
|
2005-07-22 20:48:11 +00:00
|
|
|
struct ieee80211_node *ni,
|
2009-04-26 21:34:53 +00:00
|
|
|
struct mbuf *m,
|
2008-10-26 00:43:11 +00:00
|
|
|
int type, int tid,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
const uint8_t sa[IEEE80211_ADDR_LEN],
|
|
|
|
const uint8_t da[IEEE80211_ADDR_LEN],
|
|
|
|
const uint8_t bssid[IEEE80211_ADDR_LEN])
|
2005-07-22 20:48:11 +00:00
|
|
|
{
|
|
|
|
#define WH4(wh) ((struct ieee80211_frame_addr4 *)wh)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
2011-06-04 14:28:09 +00:00
|
|
|
struct ieee80211_tx_ampdu *tap;
|
2009-04-26 21:34:53 +00:00
|
|
|
struct ieee80211_frame *wh = mtod(m, struct ieee80211_frame *);
|
|
|
|
ieee80211_seq seqno;
|
2005-07-22 20:48:11 +00:00
|
|
|
|
2013-03-09 15:35:31 +00:00
|
|
|
IEEE80211_TX_LOCK_ASSERT(ni->ni_ic);
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
|
2005-07-22 20:48:11 +00:00
|
|
|
wh->i_fc[0] = IEEE80211_FC0_VERSION_0 | type;
|
|
|
|
if ((type & IEEE80211_FC0_TYPE_MASK) == IEEE80211_FC0_TYPE_DATA) {
|
2008-04-20 20:35:46 +00:00
|
|
|
switch (vap->iv_opmode) {
|
2005-07-22 20:48:11 +00:00
|
|
|
case IEEE80211_M_STA:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_TODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, bssid);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, sa);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, da);
|
|
|
|
break;
|
|
|
|
case IEEE80211_M_IBSS:
|
|
|
|
case IEEE80211_M_AHDEMO:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, da);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, sa);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, bssid);
|
|
|
|
break;
|
|
|
|
case IEEE80211_M_HOSTAP:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_FROMDS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, da);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, bssid);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, sa);
|
|
|
|
break;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
case IEEE80211_M_WDS:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_DSTODS;
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, da);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, vap->iv_myaddr);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, da);
|
|
|
|
IEEE80211_ADDR_COPY(WH4(wh)->i_addr4, sa);
|
|
|
|
break;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
case IEEE80211_M_MBSS:
|
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (IEEE80211_IS_MULTICAST(da)) {
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_FROMDS;
|
|
|
|
/* XXX next hop */
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, da);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2,
|
|
|
|
vap->iv_myaddr);
|
|
|
|
} else {
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_DSTODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, da);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2,
|
|
|
|
vap->iv_myaddr);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, da);
|
|
|
|
IEEE80211_ADDR_COPY(WH4(wh)->i_addr4, sa);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
break;
|
2005-07-22 20:48:11 +00:00
|
|
|
case IEEE80211_M_MONITOR: /* NB: to quiet compiler */
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, da);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, sa);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS)
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, sa);
|
|
|
|
else
|
|
|
|
#endif
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, bssid);
|
2005-07-22 20:48:11 +00:00
|
|
|
}
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)&wh->i_dur[0] = 0;
|
2009-04-26 21:34:53 +00:00
|
|
|
|
2012-04-15 20:29:39 +00:00
|
|
|
tap = &ni->ni_tx_ampdu[tid];
|
2011-06-04 14:28:09 +00:00
|
|
|
if (tid != IEEE80211_NONQOS_TID && IEEE80211_AMPDU_RUNNING(tap))
|
|
|
|
m->m_flags |= M_AMPDU_MPDU;
|
|
|
|
else {
|
|
|
|
seqno = ni->ni_txseqs[tid]++;
|
|
|
|
*(uint16_t *)&wh->i_seq[0] =
|
|
|
|
htole16(seqno << IEEE80211_SEQ_SEQ_SHIFT);
|
|
|
|
M_SEQNO_SET(m, seqno);
|
|
|
|
}
|
2009-04-26 21:34:53 +00:00
|
|
|
|
|
|
|
if (IEEE80211_IS_MULTICAST(wh->i_addr1))
|
|
|
|
m->m_flags |= M_MCAST;
|
2005-07-22 20:48:11 +00:00
|
|
|
#undef WH4
|
|
|
|
}
|
|
|
|
|
2003-08-19 22:17:04 +00:00
|
|
|
/*
|
|
|
|
* Send a management frame to the specified node. The node pointer
|
|
|
|
* must have a reference as the pointer will be passed to the driver
|
|
|
|
* and potentially held for a long time. If the frame is successfully
|
|
|
|
* dispatched to the driver, then it is responsible for freeing the
|
2008-04-20 20:35:46 +00:00
|
|
|
* reference (and potentially free'ing up any associated storage);
|
|
|
|
* otherwise deal with reclaiming any reference (on error).
|
2003-08-19 22:17:04 +00:00
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
int
|
2008-10-26 00:43:11 +00:00
|
|
|
ieee80211_mgmt_output(struct ieee80211_node *ni, struct mbuf *m, int type,
|
|
|
|
struct ieee80211_bpf_params *params)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
2003-06-23 16:55:01 +00:00
|
|
|
struct ieee80211_frame *wh;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
int ret;
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2003-08-19 22:17:04 +00:00
|
|
|
KASSERT(ni != NULL, ("null node"));
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_state == IEEE80211_S_CAC) {
|
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_OUTPUT | IEEE80211_MSG_DOTH,
|
|
|
|
ni, "block %s frame in CAC state",
|
|
|
|
ieee80211_mgt_subtype_name[
|
|
|
|
(type & IEEE80211_FC0_SUBTYPE_MASK) >>
|
|
|
|
IEEE80211_FC0_SUBTYPE_SHIFT]);
|
|
|
|
vap->iv_stats.is_tx_badstate++;
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
m_freem(m);
|
|
|
|
return EIO; /* XXX */
|
|
|
|
}
|
|
|
|
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, sizeof(struct ieee80211_frame), M_NOWAIT);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (m == NULL) {
|
|
|
|
ieee80211_free_node(ni);
|
2003-06-23 16:55:01 +00:00
|
|
|
return ENOMEM;
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
2003-08-19 22:17:04 +00:00
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
IEEE80211_TX_LOCK(ic);
|
|
|
|
|
2003-06-23 16:55:01 +00:00
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
2009-04-26 21:34:53 +00:00
|
|
|
ieee80211_send_setup(ni, m,
|
2008-10-26 00:43:11 +00:00
|
|
|
IEEE80211_FC0_TYPE_MGT | type, IEEE80211_NONQOS_TID,
|
|
|
|
vap->iv_myaddr, ni->ni_macaddr, ni->ni_bssid);
|
|
|
|
if (params->ibp_flags & IEEE80211_BPF_CRYPTO) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_NOTE_MAC(vap, IEEE80211_MSG_AUTH, wh->i_addr1,
|
|
|
|
"encrypting frame (%s)", __func__);
|
2004-12-08 17:26:47 +00:00
|
|
|
wh->i_fc[1] |= IEEE80211_FC1_WEP;
|
|
|
|
}
|
2008-10-26 00:55:38 +00:00
|
|
|
m->m_flags |= M_ENCAP; /* mark encapsulated */
|
2008-10-26 00:43:11 +00:00
|
|
|
|
|
|
|
KASSERT(type != IEEE80211_FC0_SUBTYPE_PROBE_RESP, ("probe response?"));
|
|
|
|
M_WME_SETAC(m, params->ibp_pri);
|
|
|
|
|
2003-06-23 16:55:01 +00:00
|
|
|
#ifdef IEEE80211_DEBUG
|
2004-12-08 17:26:47 +00:00
|
|
|
/* avoid printing too many frames */
|
2008-04-20 20:35:46 +00:00
|
|
|
if ((ieee80211_msg_debug(vap) && doprint(vap, type)) ||
|
|
|
|
ieee80211_msg_dumppkts(vap)) {
|
2004-12-08 17:26:47 +00:00
|
|
|
printf("[%s] send %s on channel %u\n",
|
|
|
|
ether_sprintf(wh->i_addr1),
|
|
|
|
ieee80211_mgt_subtype_name[
|
|
|
|
(type & IEEE80211_FC0_SUBTYPE_MASK) >>
|
|
|
|
IEEE80211_FC0_SUBTYPE_SHIFT],
|
2005-08-10 16:22:30 +00:00
|
|
|
ieee80211_chan2ieee(ic, ic->ic_curchan));
|
2003-06-23 16:55:01 +00:00
|
|
|
}
|
2004-12-08 17:26:47 +00:00
|
|
|
#endif
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_mgmt);
|
2006-07-26 03:15:16 +00:00
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
ret = ieee80211_raw_output(vap, ni, m, params);
|
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
return (ret);
|
2006-07-26 03:15:16 +00:00
|
|
|
}
|
|
|
|
|
2003-08-19 22:17:04 +00:00
|
|
|
/*
|
2008-10-26 00:46:17 +00:00
|
|
|
* Send a null data frame to the specified node. If the station
|
|
|
|
* is setup for QoS then a QoS Null Data frame is constructed.
|
|
|
|
* If this is a WDS station then a 4-address frame is constructed.
|
2005-07-31 06:12:32 +00:00
|
|
|
*
|
|
|
|
* NB: the caller is assumed to have setup a node reference
|
|
|
|
* for use; this is necessary to deal with a race condition
|
2008-04-20 20:35:46 +00:00
|
|
|
* when probing for inactive stations. Like ieee80211_mgmt_output
|
|
|
|
* we must cleanup any node reference on error; however we
|
|
|
|
* can safely just unref it as we know it will never be the
|
|
|
|
* last reference to the node.
|
2004-12-08 17:26:47 +00:00
|
|
|
*/
|
|
|
|
int
|
2005-07-22 17:31:18 +00:00
|
|
|
ieee80211_send_nulldata(struct ieee80211_node *ni)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
2005-07-22 17:31:18 +00:00
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
2004-12-08 17:26:47 +00:00
|
|
|
struct mbuf *m;
|
|
|
|
struct ieee80211_frame *wh;
|
2008-10-26 00:46:17 +00:00
|
|
|
int hdrlen;
|
|
|
|
uint8_t *frm;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
int ret;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_state == IEEE80211_S_CAC) {
|
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_OUTPUT | IEEE80211_MSG_DOTH,
|
|
|
|
ni, "block %s frame in CAC state", "null data");
|
|
|
|
ieee80211_unref_node(&ni);
|
|
|
|
vap->iv_stats.is_tx_badstate++;
|
|
|
|
return EIO; /* XXX */
|
|
|
|
}
|
|
|
|
|
2008-10-26 00:46:17 +00:00
|
|
|
if (ni->ni_flags & (IEEE80211_NODE_QOS|IEEE80211_NODE_HT))
|
|
|
|
hdrlen = sizeof(struct ieee80211_qosframe);
|
|
|
|
else
|
|
|
|
hdrlen = sizeof(struct ieee80211_frame);
|
|
|
|
/* NB: only WDS vap's get 4-address frames */
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_WDS)
|
|
|
|
hdrlen += IEEE80211_ADDR_LEN;
|
|
|
|
if (ic->ic_flags & IEEE80211_F_DATAPAD)
|
|
|
|
hdrlen = roundup(hdrlen, sizeof(uint32_t));
|
|
|
|
|
|
|
|
m = ieee80211_getmgtframe(&frm, ic->ic_headroom + hdrlen, 0);
|
2004-12-08 17:26:47 +00:00
|
|
|
if (m == NULL) {
|
|
|
|
/* XXX debug msg */
|
2005-07-31 06:12:32 +00:00
|
|
|
ieee80211_unref_node(&ni);
|
2008-04-20 20:35:46 +00:00
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
2004-12-08 17:26:47 +00:00
|
|
|
return ENOMEM;
|
|
|
|
}
|
2008-10-26 00:46:17 +00:00
|
|
|
KASSERT(M_LEADINGSPACE(m) >= hdrlen,
|
|
|
|
("leading space %zd", M_LEADINGSPACE(m)));
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, hdrlen, M_NOWAIT);
|
2008-10-26 00:46:17 +00:00
|
|
|
if (m == NULL) {
|
|
|
|
/* NB: cannot happen */
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
return ENOMEM;
|
|
|
|
}
|
2004-12-08 17:26:47 +00:00
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
IEEE80211_TX_LOCK(ic);
|
|
|
|
|
2008-10-26 00:46:17 +00:00
|
|
|
wh = mtod(m, struct ieee80211_frame *); /* NB: a little lie */
|
|
|
|
if (ni->ni_flags & IEEE80211_NODE_QOS) {
|
|
|
|
const int tid = WME_AC_TO_TID(WME_AC_BE);
|
|
|
|
uint8_t *qos;
|
|
|
|
|
2009-04-26 21:34:53 +00:00
|
|
|
ieee80211_send_setup(ni, m,
|
2008-10-26 00:46:17 +00:00
|
|
|
IEEE80211_FC0_TYPE_DATA | IEEE80211_FC0_SUBTYPE_QOS_NULL,
|
|
|
|
tid, vap->iv_myaddr, ni->ni_macaddr, ni->ni_bssid);
|
|
|
|
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_WDS)
|
|
|
|
qos = ((struct ieee80211_qosframe_addr4 *) wh)->i_qos;
|
|
|
|
else
|
|
|
|
qos = ((struct ieee80211_qosframe *) wh)->i_qos;
|
|
|
|
qos[0] = tid & IEEE80211_QOS_TID;
|
|
|
|
if (ic->ic_wme.wme_wmeChanParams.cap_wmeParams[WME_AC_BE].wmep_noackPolicy)
|
|
|
|
qos[0] |= IEEE80211_QOS_ACKPOLICY_NOACK;
|
|
|
|
qos[1] = 0;
|
|
|
|
} else {
|
2009-04-26 21:34:53 +00:00
|
|
|
ieee80211_send_setup(ni, m,
|
2008-10-26 00:46:17 +00:00
|
|
|
IEEE80211_FC0_TYPE_DATA | IEEE80211_FC0_SUBTYPE_NODATA,
|
|
|
|
IEEE80211_NONQOS_TID,
|
|
|
|
vap->iv_myaddr, ni->ni_macaddr, ni->ni_bssid);
|
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_opmode != IEEE80211_M_WDS) {
|
|
|
|
/* NB: power management bit is never sent by an AP */
|
|
|
|
if ((ni->ni_flags & IEEE80211_NODE_PWR_MGT) &&
|
|
|
|
vap->iv_opmode != IEEE80211_M_HOSTAP)
|
|
|
|
wh->i_fc[1] |= IEEE80211_FC1_PWR_MGT;
|
|
|
|
}
|
2008-10-26 00:46:17 +00:00
|
|
|
m->m_len = m->m_pkthdr.len = hdrlen;
|
2008-10-26 00:55:38 +00:00
|
|
|
m->m_flags |= M_ENCAP; /* mark encapsulated */
|
2008-10-26 00:46:17 +00:00
|
|
|
|
2007-09-18 21:07:41 +00:00
|
|
|
M_WME_SETAC(m, WME_AC_BE);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_data);
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_DEBUG | IEEE80211_MSG_DUMPPKTS, ni,
|
2008-10-26 00:46:17 +00:00
|
|
|
"send %snull data frame on channel %u, pwr mgt %s",
|
|
|
|
ni->ni_flags & IEEE80211_NODE_QOS ? "QoS " : "",
|
2005-08-10 16:22:30 +00:00
|
|
|
ieee80211_chan2ieee(ic, ic->ic_curchan),
|
2005-07-22 20:48:11 +00:00
|
|
|
wh->i_fc[1] & IEEE80211_FC1_PWR_MGT ? "ena" : "dis");
|
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
ret = ieee80211_raw_output(vap, ni, m, NULL);
|
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
return (ret);
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Assign priority to a frame based on any vlan tag assigned
|
|
|
|
* to the station and/or any Diffserv setting in an IP header.
|
|
|
|
* Finally, if an ACM policy is setup (in station mode) it's
|
|
|
|
* applied.
|
|
|
|
*/
|
|
|
|
int
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_classify(struct ieee80211_node *ni, struct mbuf *m)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
const struct ether_header *eh = mtod(m, struct ether_header *);
|
2004-12-08 17:26:47 +00:00
|
|
|
int v_wme_ac, d_wme_ac, ac;
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Always promote PAE/EAPOL frames to high priority.
|
|
|
|
*/
|
|
|
|
if (eh->ether_type == htons(ETHERTYPE_PAE)) {
|
|
|
|
/* NB: mark so others don't need to check header */
|
|
|
|
m->m_flags |= M_EAPOL;
|
|
|
|
ac = WME_AC_VO;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Non-qos traffic goes to BE.
|
|
|
|
*/
|
2004-12-08 17:26:47 +00:00
|
|
|
if ((ni->ni_flags & IEEE80211_NODE_QOS) == 0) {
|
|
|
|
ac = WME_AC_BE;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If node has a vlan tag then all traffic
|
|
|
|
* to it must have a matching tag.
|
|
|
|
*/
|
|
|
|
v_wme_ac = 0;
|
|
|
|
if (ni->ni_vlan != 0) {
|
2006-09-17 13:33:30 +00:00
|
|
|
if ((m->m_flags & M_VLANTAG) == 0) {
|
2004-12-08 17:26:47 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_novlantag);
|
|
|
|
return 1;
|
|
|
|
}
|
2006-09-17 13:33:30 +00:00
|
|
|
if (EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) !=
|
2004-12-08 17:26:47 +00:00
|
|
|
EVL_VLANOFTAG(ni->ni_vlan)) {
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_vlanmismatch);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
/* map vlan priority to AC */
|
2007-11-23 06:24:57 +00:00
|
|
|
v_wme_ac = TID_TO_WME_AC(EVL_PRIOFTAG(ni->ni_vlan));
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
|
2009-06-07 22:06:15 +00:00
|
|
|
/* XXX m_copydata may be too slow for fast path */
|
2004-12-08 17:26:47 +00:00
|
|
|
#ifdef INET
|
|
|
|
if (eh->ether_type == htons(ETHERTYPE_IP)) {
|
2007-11-23 06:24:57 +00:00
|
|
|
uint8_t tos;
|
2004-12-08 17:26:47 +00:00
|
|
|
/*
|
2007-11-23 06:24:57 +00:00
|
|
|
* IP frame, map the DSCP bits from the TOS field.
|
2004-12-08 17:26:47 +00:00
|
|
|
*/
|
2007-11-23 06:24:57 +00:00
|
|
|
/* NB: ip header may not be in first mbuf */
|
|
|
|
m_copydata(m, sizeof(struct ether_header) +
|
|
|
|
offsetof(struct ip, ip_tos), sizeof(tos), &tos);
|
|
|
|
tos >>= 5; /* NB: ECN + low 3 bits of DSCP */
|
|
|
|
d_wme_ac = TID_TO_WME_AC(tos);
|
2004-12-08 17:26:47 +00:00
|
|
|
} else {
|
|
|
|
#endif /* INET */
|
2009-06-07 22:06:15 +00:00
|
|
|
#ifdef INET6
|
|
|
|
if (eh->ether_type == htons(ETHERTYPE_IPV6)) {
|
|
|
|
uint32_t flow;
|
|
|
|
uint8_t tos;
|
|
|
|
/*
|
2012-01-07 23:41:38 +00:00
|
|
|
* IPv6 frame, map the DSCP bits from the traffic class field.
|
2009-06-07 22:06:15 +00:00
|
|
|
*/
|
|
|
|
m_copydata(m, sizeof(struct ether_header) +
|
|
|
|
offsetof(struct ip6_hdr, ip6_flow), sizeof(flow),
|
|
|
|
(caddr_t) &flow);
|
|
|
|
tos = (uint8_t)(ntohl(flow) >> 20);
|
|
|
|
tos >>= 5; /* NB: ECN + low 3 bits of DSCP */
|
|
|
|
d_wme_ac = TID_TO_WME_AC(tos);
|
|
|
|
} else {
|
|
|
|
#endif /* INET6 */
|
2004-12-08 17:26:47 +00:00
|
|
|
d_wme_ac = WME_AC_BE;
|
2009-06-07 22:06:15 +00:00
|
|
|
#ifdef INET6
|
|
|
|
}
|
|
|
|
#endif
|
2004-12-08 17:26:47 +00:00
|
|
|
#ifdef INET
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
/*
|
|
|
|
* Use highest priority AC.
|
|
|
|
*/
|
|
|
|
if (v_wme_ac > d_wme_ac)
|
|
|
|
ac = v_wme_ac;
|
|
|
|
else
|
|
|
|
ac = d_wme_ac;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Apply ACM policy.
|
|
|
|
*/
|
2008-04-20 20:35:46 +00:00
|
|
|
if (ni->ni_vap->iv_opmode == IEEE80211_M_STA) {
|
2004-12-08 17:26:47 +00:00
|
|
|
static const int acmap[4] = {
|
|
|
|
WME_AC_BK, /* WME_AC_BE */
|
|
|
|
WME_AC_BK, /* WME_AC_BK */
|
|
|
|
WME_AC_BE, /* WME_AC_VI */
|
|
|
|
WME_AC_VI, /* WME_AC_VO */
|
|
|
|
};
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
while (ac != WME_AC_BK &&
|
|
|
|
ic->ic_wme.wme_wmeBssChanParams.cap_wmeParams[ac].wmep_acm)
|
|
|
|
ac = acmap[ac];
|
|
|
|
}
|
|
|
|
done:
|
|
|
|
M_WME_SETAC(m, ac);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2004-12-31 21:54:53 +00:00
|
|
|
* Insure there is sufficient contiguous space to encapsulate the
|
|
|
|
* 802.11 data frame. If room isn't already there, arrange for it.
|
|
|
|
* Drivers and cipher modules assume we have done the necessary work
|
|
|
|
* and fail rudely if they don't find the space they need.
|
|
|
|
*/
|
2009-03-24 20:39:08 +00:00
|
|
|
struct mbuf *
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_mbuf_adjust(struct ieee80211vap *vap, int hdrsize,
|
2004-12-31 21:54:53 +00:00
|
|
|
struct ieee80211_key *key, struct mbuf *m)
|
|
|
|
{
|
2006-12-01 17:58:33 +00:00
|
|
|
#define TO_BE_RECLAIMED (sizeof(struct ether_header) - sizeof(struct llc))
|
2008-04-20 20:35:46 +00:00
|
|
|
int needed_space = vap->iv_ic->ic_headroom + hdrsize;
|
2004-12-31 21:54:53 +00:00
|
|
|
|
|
|
|
if (key != NULL) {
|
|
|
|
/* XXX belongs in crypto code? */
|
|
|
|
needed_space += key->wk_cipher->ic_header;
|
|
|
|
/* XXX frags */
|
2006-03-15 21:27:08 +00:00
|
|
|
/*
|
|
|
|
* When crypto is being done in the host we must insure
|
|
|
|
* the data are writable for the cipher routines; clone
|
|
|
|
* a writable mbuf chain.
|
|
|
|
* XXX handle SWMIC specially
|
|
|
|
*/
|
2008-05-28 23:25:36 +00:00
|
|
|
if (key->wk_flags & (IEEE80211_KEY_SWENCRYPT|IEEE80211_KEY_SWENMIC)) {
|
2006-03-15 21:27:08 +00:00
|
|
|
m = m_unshare(m, M_NOWAIT);
|
|
|
|
if (m == NULL) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_OUTPUT,
|
2006-03-15 21:27:08 +00:00
|
|
|
"%s: cannot get writable mbuf\n", __func__);
|
2008-04-20 20:35:46 +00:00
|
|
|
vap->iv_stats.is_tx_nobuf++; /* XXX new stat */
|
2006-03-15 21:27:08 +00:00
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
}
|
2004-12-31 21:54:53 +00:00
|
|
|
}
|
|
|
|
/*
|
|
|
|
* We know we are called just before stripping an Ethernet
|
|
|
|
* header and prepending an LLC header. This means we know
|
|
|
|
* there will be
|
2006-12-01 17:58:33 +00:00
|
|
|
* sizeof(struct ether_header) - sizeof(struct llc)
|
2004-12-31 21:54:53 +00:00
|
|
|
* bytes recovered to which we need additional space for the
|
|
|
|
* 802.11 header and any crypto header.
|
|
|
|
*/
|
|
|
|
/* XXX check trailing space and copy instead? */
|
|
|
|
if (M_LEADINGSPACE(m) < needed_space - TO_BE_RECLAIMED) {
|
|
|
|
struct mbuf *n = m_gethdr(M_NOWAIT, m->m_type);
|
|
|
|
if (n == NULL) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_OUTPUT,
|
2004-12-31 21:54:53 +00:00
|
|
|
"%s: cannot expand storage\n", __func__);
|
2008-04-20 20:35:46 +00:00
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
2004-12-31 21:54:53 +00:00
|
|
|
m_freem(m);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
KASSERT(needed_space <= MHLEN,
|
2013-07-07 21:39:37 +00:00
|
|
|
("not enough room, need %u got %d\n", needed_space, MHLEN));
|
2004-12-31 21:54:53 +00:00
|
|
|
/*
|
|
|
|
* Setup new mbuf to have leading space to prepend the
|
|
|
|
* 802.11 header and any crypto header bits that are
|
|
|
|
* required (the latter are added when the driver calls
|
|
|
|
* back to ieee80211_crypto_encap to do crypto encapsulation).
|
|
|
|
*/
|
|
|
|
/* NB: must be first 'cuz it clobbers m_data */
|
|
|
|
m_move_pkthdr(n, m);
|
|
|
|
n->m_len = 0; /* NB: m_gethdr does not set */
|
|
|
|
n->m_data += needed_space;
|
|
|
|
/*
|
|
|
|
* Pull up Ethernet header to create the expected layout.
|
|
|
|
* We could use m_pullup but that's overkill (i.e. we don't
|
|
|
|
* need the actual data) and it cannot fail so do it inline
|
|
|
|
* for speed.
|
|
|
|
*/
|
|
|
|
/* NB: struct ether_header is known to be contiguous */
|
|
|
|
n->m_len += sizeof(struct ether_header);
|
|
|
|
m->m_len -= sizeof(struct ether_header);
|
|
|
|
m->m_data += sizeof(struct ether_header);
|
|
|
|
/*
|
|
|
|
* Replace the head of the chain.
|
|
|
|
*/
|
|
|
|
n->m_next = m;
|
|
|
|
m = n;
|
|
|
|
}
|
|
|
|
return m;
|
|
|
|
#undef TO_BE_RECLAIMED
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Return the transmit key to use in sending a unicast frame.
|
|
|
|
* If a unicast key is set we use that. When no unicast key is set
|
|
|
|
* we fall back to the default transmit key.
|
2004-12-08 17:26:47 +00:00
|
|
|
*/
|
|
|
|
static __inline struct ieee80211_key *
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_crypto_getucastkey(struct ieee80211vap *vap,
|
|
|
|
struct ieee80211_node *ni)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
2007-03-11 06:36:10 +00:00
|
|
|
if (IEEE80211_KEY_UNDEFINED(&ni->ni_ucastkey)) {
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_def_txkey == IEEE80211_KEYIX_NONE ||
|
|
|
|
IEEE80211_KEY_UNDEFINED(&vap->iv_nw_keys[vap->iv_def_txkey]))
|
2004-12-08 17:26:47 +00:00
|
|
|
return NULL;
|
2008-04-20 20:35:46 +00:00
|
|
|
return &vap->iv_nw_keys[vap->iv_def_txkey];
|
2004-12-08 17:26:47 +00:00
|
|
|
} else {
|
|
|
|
return &ni->ni_ucastkey;
|
|
|
|
}
|
2004-12-31 21:54:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Return the transmit key to use in sending a multicast frame.
|
|
|
|
* Multicast traffic always uses the group key which is installed as
|
|
|
|
* the default tx key.
|
|
|
|
*/
|
|
|
|
static __inline struct ieee80211_key *
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_crypto_getmcastkey(struct ieee80211vap *vap,
|
|
|
|
struct ieee80211_node *ni)
|
2004-12-31 21:54:53 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_def_txkey == IEEE80211_KEYIX_NONE ||
|
|
|
|
IEEE80211_KEY_UNDEFINED(&vap->iv_nw_keys[vap->iv_def_txkey]))
|
2004-12-31 21:54:53 +00:00
|
|
|
return NULL;
|
2008-04-20 20:35:46 +00:00
|
|
|
return &vap->iv_nw_keys[vap->iv_def_txkey];
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Encapsulate an outbound data frame. The mbuf chain is updated.
|
|
|
|
* If an error is encountered NULL is returned. The caller is required
|
|
|
|
* to provide a node reference and pullup the ethernet header in the
|
|
|
|
* first mbuf.
|
2008-04-20 20:35:46 +00:00
|
|
|
*
|
|
|
|
* NB: Packet is assumed to be processed by ieee80211_classify which
|
|
|
|
* marked EAPOL frames w/ M_EAPOL.
|
2003-08-19 22:17:04 +00:00
|
|
|
*/
|
2003-06-23 16:55:01 +00:00
|
|
|
struct mbuf *
|
2009-03-30 21:53:27 +00:00
|
|
|
ieee80211_encap(struct ieee80211vap *vap, struct ieee80211_node *ni,
|
|
|
|
struct mbuf *m)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
#define WH4(wh) ((struct ieee80211_frame_addr4 *)(wh))
|
2012-05-01 15:56:26 +00:00
|
|
|
#define MC01(mc) ((struct ieee80211_meshcntl_ae01 *)mc)
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
struct ieee80211_mesh_state *ms = vap->iv_mesh;
|
2009-07-20 19:12:08 +00:00
|
|
|
struct ieee80211_meshcntl_ae10 *mc;
|
2012-05-01 15:56:26 +00:00
|
|
|
struct ieee80211_mesh_route *rt = NULL;
|
|
|
|
int dir = -1;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#endif
|
2003-06-23 16:55:01 +00:00
|
|
|
struct ether_header eh;
|
|
|
|
struct ieee80211_frame *wh;
|
2004-12-08 17:26:47 +00:00
|
|
|
struct ieee80211_key *key;
|
2003-06-23 16:55:01 +00:00
|
|
|
struct llc *llc;
|
2009-03-24 20:39:08 +00:00
|
|
|
int hdrsize, hdrspace, datalen, addqos, txfrag, is4addr;
|
2009-04-26 21:34:53 +00:00
|
|
|
ieee80211_seq seqno;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
int meshhdrsize, meshae;
|
|
|
|
uint8_t *qos;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
|
|
|
|
IEEE80211_TX_LOCK_ASSERT(ic);
|
2003-06-23 16:55:01 +00:00
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
/*
|
|
|
|
* Copy existing Ethernet header to a safe place. The
|
|
|
|
* rest of the code assumes it's ok to strip it when
|
|
|
|
* reorganizing state for the final encapsulation.
|
|
|
|
*/
|
2004-12-08 17:26:47 +00:00
|
|
|
KASSERT(m->m_len >= sizeof(eh), ("no ethernet header!"));
|
2008-04-20 20:35:46 +00:00
|
|
|
ETHER_HEADER_COPY(&eh, mtod(m, caddr_t));
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
/*
|
|
|
|
* Insure space for additional headers. First identify
|
|
|
|
* transmit key to use in calculating any buffer adjustments
|
|
|
|
* required. This is also used below to do privacy
|
|
|
|
* encapsulation work. Then calculate the 802.11 header
|
|
|
|
* size and any padding required by the driver.
|
|
|
|
*
|
|
|
|
* Note key may be NULL if we fall back to the default
|
|
|
|
* transmit key and that is not set. In that case the
|
|
|
|
* buffer may not be expanded as needed by the cipher
|
|
|
|
* routines, but they will/should discard it.
|
|
|
|
*/
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_PRIVACY) {
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_STA ||
|
|
|
|
!IEEE80211_IS_MULTICAST(eh.ether_dhost) ||
|
|
|
|
(vap->iv_opmode == IEEE80211_M_WDS &&
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_WDSLEGACY)))
|
|
|
|
key = ieee80211_crypto_getucastkey(vap, ni);
|
2004-12-31 21:54:53 +00:00
|
|
|
else
|
2008-04-20 20:35:46 +00:00
|
|
|
key = ieee80211_crypto_getmcastkey(vap, ni);
|
|
|
|
if (key == NULL && (m->m_flags & M_EAPOL) == 0) {
|
|
|
|
IEEE80211_NOTE_MAC(vap, IEEE80211_MSG_CRYPTO,
|
|
|
|
eh.ether_dhost,
|
|
|
|
"no default transmit key (%s) deftxkey %u",
|
|
|
|
__func__, vap->iv_def_txkey);
|
|
|
|
vap->iv_stats.is_tx_nodefkey++;
|
2007-08-24 15:44:27 +00:00
|
|
|
goto bad;
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
} else
|
|
|
|
key = NULL;
|
2004-12-31 21:54:53 +00:00
|
|
|
/*
|
|
|
|
* XXX Some ap's don't handle QoS-encapsulated EAPOL
|
|
|
|
* frames so suppress use. This may be an issue if other
|
|
|
|
* ap's require all data frames to be QoS-encapsulated
|
|
|
|
* once negotiated in which case we'll need to make this
|
|
|
|
* configurable.
|
2012-03-04 05:52:26 +00:00
|
|
|
* NB: mesh data frames are QoS.
|
2004-12-31 21:54:53 +00:00
|
|
|
*/
|
2012-03-04 05:52:26 +00:00
|
|
|
addqos = ((ni->ni_flags & (IEEE80211_NODE_QOS|IEEE80211_NODE_HT)) ||
|
|
|
|
(vap->iv_opmode == IEEE80211_M_MBSS)) &&
|
|
|
|
(m->m_flags & M_EAPOL) == 0;
|
2004-12-31 21:54:53 +00:00
|
|
|
if (addqos)
|
2004-12-08 17:26:47 +00:00
|
|
|
hdrsize = sizeof(struct ieee80211_qosframe);
|
|
|
|
else
|
|
|
|
hdrsize = sizeof(struct ieee80211_frame);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS) {
|
|
|
|
/*
|
|
|
|
* Mesh data frames are encapsulated according to the
|
|
|
|
* rules of Section 11B.8.5 (p.139 of D3.0 spec).
|
|
|
|
* o Group Addressed data (aka multicast) originating
|
|
|
|
* at the local sta are sent w/ 3-address format and
|
|
|
|
* address extension mode 00
|
|
|
|
* o Individually Addressed data (aka unicast) originating
|
|
|
|
* at the local sta are sent w/ 4-address format and
|
|
|
|
* address extension mode 00
|
|
|
|
* o Group Addressed data forwarded from a non-mesh sta are
|
|
|
|
* sent w/ 3-address format and address extension mode 01
|
|
|
|
* o Individually Address data from another sta are sent
|
|
|
|
* w/ 4-address format and address extension mode 10
|
|
|
|
*/
|
|
|
|
is4addr = 0; /* NB: don't use, disable */
|
2012-05-01 15:56:26 +00:00
|
|
|
if (!IEEE80211_IS_MULTICAST(eh.ether_dhost)) {
|
|
|
|
rt = ieee80211_mesh_rt_find(vap, eh.ether_dhost);
|
|
|
|
KASSERT(rt != NULL, ("route is NULL"));
|
|
|
|
dir = IEEE80211_FC1_DIR_DSTODS;
|
|
|
|
hdrsize += IEEE80211_ADDR_LEN;
|
|
|
|
if (rt->rt_flags & IEEE80211_MESHRT_FLAGS_PROXY) {
|
|
|
|
if (IEEE80211_ADDR_EQ(rt->rt_mesh_gate,
|
|
|
|
vap->iv_myaddr)) {
|
|
|
|
IEEE80211_NOTE_MAC(vap,
|
|
|
|
IEEE80211_MSG_MESH,
|
|
|
|
eh.ether_dhost,
|
|
|
|
"%s", "trying to send to ourself");
|
|
|
|
goto bad;
|
|
|
|
}
|
|
|
|
meshae = IEEE80211_MESH_AE_10;
|
|
|
|
meshhdrsize =
|
|
|
|
sizeof(struct ieee80211_meshcntl_ae10);
|
|
|
|
} else {
|
|
|
|
meshae = IEEE80211_MESH_AE_00;
|
|
|
|
meshhdrsize =
|
|
|
|
sizeof(struct ieee80211_meshcntl);
|
|
|
|
}
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
} else {
|
2012-05-01 15:56:26 +00:00
|
|
|
dir = IEEE80211_FC1_DIR_FROMDS;
|
|
|
|
if (!IEEE80211_ADDR_EQ(eh.ether_shost, vap->iv_myaddr)) {
|
|
|
|
/* proxy group */
|
|
|
|
meshae = IEEE80211_MESH_AE_01;
|
|
|
|
meshhdrsize =
|
|
|
|
sizeof(struct ieee80211_meshcntl_ae01);
|
|
|
|
} else {
|
|
|
|
/* group */
|
|
|
|
meshae = IEEE80211_MESH_AE_00;
|
|
|
|
meshhdrsize = sizeof(struct ieee80211_meshcntl);
|
|
|
|
}
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
}
|
|
|
|
} else {
|
|
|
|
#endif
|
|
|
|
/*
|
|
|
|
* 4-address frames need to be generated for:
|
|
|
|
* o packets sent through a WDS vap (IEEE80211_M_WDS)
|
|
|
|
* o packets sent through a vap marked for relaying
|
|
|
|
* (e.g. a station operating with dynamic WDS)
|
|
|
|
*/
|
|
|
|
is4addr = vap->iv_opmode == IEEE80211_M_WDS ||
|
|
|
|
((vap->iv_flags_ext & IEEE80211_FEXT_4ADDR) &&
|
|
|
|
!IEEE80211_ADDR_EQ(eh.ether_shost, vap->iv_myaddr));
|
|
|
|
if (is4addr)
|
|
|
|
hdrsize += IEEE80211_ADDR_LEN;
|
|
|
|
meshhdrsize = meshae = 0;
|
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
}
|
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Honor driver DATAPAD requirement.
|
|
|
|
*/
|
2004-12-08 17:26:47 +00:00
|
|
|
if (ic->ic_flags & IEEE80211_F_DATAPAD)
|
2008-04-20 20:35:46 +00:00
|
|
|
hdrspace = roundup(hdrsize, sizeof(uint32_t));
|
|
|
|
else
|
|
|
|
hdrspace = hdrsize;
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2009-03-24 20:39:08 +00:00
|
|
|
if (__predict_true((m->m_flags & M_FF) == 0)) {
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
/*
|
|
|
|
* Normal frame.
|
|
|
|
*/
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
m = ieee80211_mbuf_adjust(vap, hdrspace + meshhdrsize, key, m);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
if (m == NULL) {
|
|
|
|
/* NB: ieee80211_mbuf_adjust handles msgs+statistics */
|
|
|
|
goto bad;
|
|
|
|
}
|
|
|
|
/* NB: this could be optimized 'cuz of ieee80211_mbuf_adjust */
|
|
|
|
m_adj(m, sizeof(struct ether_header) - sizeof(struct llc));
|
|
|
|
llc = mtod(m, struct llc *);
|
|
|
|
llc->llc_dsap = llc->llc_ssap = LLC_SNAP_LSAP;
|
|
|
|
llc->llc_control = LLC_UI;
|
|
|
|
llc->llc_snap.org_code[0] = 0;
|
|
|
|
llc->llc_snap.org_code[1] = 0;
|
|
|
|
llc->llc_snap.org_code[2] = 0;
|
|
|
|
llc->llc_snap.ether_type = eh.ether_type;
|
2009-03-24 20:39:08 +00:00
|
|
|
} else {
|
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2009-03-30 21:53:27 +00:00
|
|
|
/*
|
|
|
|
* Aggregated frame.
|
|
|
|
*/
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
m = ieee80211_ff_encap(vap, m, hdrspace + meshhdrsize, key);
|
2009-03-24 20:39:08 +00:00
|
|
|
if (m == NULL)
|
|
|
|
#endif
|
|
|
|
goto bad;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
}
|
2004-12-08 17:26:47 +00:00
|
|
|
datalen = m->m_pkthdr.len; /* NB: w/o 802.11 header */
|
|
|
|
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, hdrspace + meshhdrsize, M_NOWAIT);
|
2003-10-17 23:15:30 +00:00
|
|
|
if (m == NULL) {
|
2008-04-20 20:35:46 +00:00
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
2003-08-19 22:17:04 +00:00
|
|
|
goto bad;
|
2003-10-17 23:15:30 +00:00
|
|
|
}
|
2003-06-23 16:55:01 +00:00
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
|
|
|
wh->i_fc[0] = IEEE80211_FC0_VERSION_0 | IEEE80211_FC0_TYPE_DATA;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)wh->i_dur = 0;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
qos = NULL; /* NB: quiet compiler */
|
2008-04-20 20:35:46 +00:00
|
|
|
if (is4addr) {
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_DSTODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, ni->ni_macaddr);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, vap->iv_myaddr);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(WH4(wh)->i_addr4, eh.ether_shost);
|
|
|
|
} else switch (vap->iv_opmode) {
|
2003-06-23 16:55:01 +00:00
|
|
|
case IEEE80211_M_STA:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_TODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, ni->ni_bssid);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, eh.ether_shost);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, eh.ether_dhost);
|
|
|
|
break;
|
|
|
|
case IEEE80211_M_IBSS:
|
|
|
|
case IEEE80211_M_AHDEMO:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, eh.ether_shost);
|
2005-01-22 20:33:37 +00:00
|
|
|
/*
|
2008-04-20 20:35:46 +00:00
|
|
|
* NB: always use the bssid from iv_bss as the
|
2005-01-22 20:33:37 +00:00
|
|
|
* neighbor's may be stale after an ibss merge
|
|
|
|
*/
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, vap->iv_bss->ni_bssid);
|
2003-06-23 16:55:01 +00:00
|
|
|
break;
|
|
|
|
case IEEE80211_M_HOSTAP:
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_FROMDS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, ni->ni_bssid);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, eh.ether_shost);
|
|
|
|
break;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
case IEEE80211_M_MBSS:
|
|
|
|
/* NB: offset by hdrspace to deal with DATAPAD */
|
2009-07-20 19:12:08 +00:00
|
|
|
mc = (struct ieee80211_meshcntl_ae10 *)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
(mtod(m, uint8_t *) + hdrspace);
|
2012-05-01 15:56:26 +00:00
|
|
|
wh->i_fc[1] = dir;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
switch (meshae) {
|
2012-05-01 15:56:26 +00:00
|
|
|
case IEEE80211_MESH_AE_00: /* no proxy */
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
mc->mc_flags = 0;
|
2012-05-01 15:56:26 +00:00
|
|
|
if (dir == IEEE80211_FC1_DIR_DSTODS) { /* ucast */
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1,
|
|
|
|
ni->ni_macaddr);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2,
|
|
|
|
vap->iv_myaddr);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3,
|
|
|
|
eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(WH4(wh)->i_addr4,
|
|
|
|
eh.ether_shost);
|
|
|
|
qos =((struct ieee80211_qosframe_addr4 *)
|
|
|
|
wh)->i_qos;
|
|
|
|
} else if (dir == IEEE80211_FC1_DIR_FROMDS) {
|
|
|
|
/* mcast */
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1,
|
|
|
|
eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2,
|
|
|
|
vap->iv_myaddr);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3,
|
|
|
|
eh.ether_shost);
|
|
|
|
qos = ((struct ieee80211_qosframe *)
|
|
|
|
wh)->i_qos;
|
|
|
|
}
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
break;
|
2012-05-01 15:56:26 +00:00
|
|
|
case IEEE80211_MESH_AE_01: /* mcast, proxy */
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_FROMDS;
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, vap->iv_myaddr);
|
2009-07-20 19:12:08 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, vap->iv_myaddr);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
mc->mc_flags = 1;
|
2012-05-01 15:56:26 +00:00
|
|
|
IEEE80211_ADDR_COPY(MC01(mc)->mc_addr4,
|
|
|
|
eh.ether_shost);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
qos = ((struct ieee80211_qosframe *) wh)->i_qos;
|
|
|
|
break;
|
2012-05-01 15:56:26 +00:00
|
|
|
case IEEE80211_MESH_AE_10: /* ucast, proxy */
|
|
|
|
KASSERT(rt != NULL, ("route is NULL"));
|
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, rt->rt_nexthop);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, vap->iv_myaddr);
|
2012-05-01 15:56:26 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, rt->rt_mesh_gate);
|
2009-07-20 19:12:08 +00:00
|
|
|
IEEE80211_ADDR_COPY(WH4(wh)->i_addr4, vap->iv_myaddr);
|
2012-05-01 15:56:26 +00:00
|
|
|
mc->mc_flags = IEEE80211_MESH_AE_10;
|
|
|
|
IEEE80211_ADDR_COPY(mc->mc_addr5, eh.ether_dhost);
|
|
|
|
IEEE80211_ADDR_COPY(mc->mc_addr6, eh.ether_shost);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
qos = ((struct ieee80211_qosframe_addr4 *) wh)->i_qos;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
KASSERT(0, ("meshae %d", meshae));
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
mc->mc_ttl = ms->ms_ttl;
|
|
|
|
ms->ms_seq++;
|
|
|
|
LE_WRITE_4(mc->mc_seq, ms->ms_seq);
|
|
|
|
break;
|
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
case IEEE80211_M_WDS: /* NB: is4addr should always be true */
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
default:
|
2003-08-19 22:17:04 +00:00
|
|
|
goto bad;
|
2003-06-23 16:55:01 +00:00
|
|
|
}
|
2009-04-03 20:46:32 +00:00
|
|
|
if (m->m_flags & M_MORE_DATA)
|
2005-07-06 01:55:17 +00:00
|
|
|
wh->i_fc[1] |= IEEE80211_FC1_MORE_DATA;
|
2004-12-31 21:54:53 +00:00
|
|
|
if (addqos) {
|
2004-12-08 17:26:47 +00:00
|
|
|
int ac, tid;
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (is4addr) {
|
|
|
|
qos = ((struct ieee80211_qosframe_addr4 *) wh)->i_qos;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
/* NB: mesh case handled earlier */
|
|
|
|
} else if (vap->iv_opmode != IEEE80211_M_MBSS)
|
2008-04-20 20:35:46 +00:00
|
|
|
qos = ((struct ieee80211_qosframe *) wh)->i_qos;
|
2004-12-08 17:26:47 +00:00
|
|
|
ac = M_WME_GETAC(m);
|
|
|
|
/* map from access class/queue to 11e header priorty value */
|
|
|
|
tid = WME_AC_TO_TID(ac);
|
2008-04-20 20:35:46 +00:00
|
|
|
qos[0] = tid & IEEE80211_QOS_TID;
|
2004-12-08 17:26:47 +00:00
|
|
|
if (ic->ic_wme.wme_wmeChanParams.cap_wmeParams[ac].wmep_noackPolicy)
|
2008-04-20 20:35:46 +00:00
|
|
|
qos[0] |= IEEE80211_QOS_ACKPOLICY_NOACK;
|
2012-03-04 05:52:26 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
2012-06-25 11:52:26 +00:00
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS)
|
|
|
|
qos[1] = IEEE80211_QOS_MC;
|
|
|
|
else
|
2012-03-04 05:52:26 +00:00
|
|
|
#endif
|
|
|
|
qos[1] = 0;
|
2008-04-20 20:35:46 +00:00
|
|
|
wh->i_fc[0] |= IEEE80211_FC0_SUBTYPE_QOS;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2008-09-21 23:00:19 +00:00
|
|
|
if ((m->m_flags & M_AMPDU_MPDU) == 0) {
|
|
|
|
/*
|
|
|
|
* NB: don't assign a sequence # to potential
|
|
|
|
* aggregates; we expect this happens at the
|
|
|
|
* point the frame comes off any aggregation q
|
|
|
|
* as otherwise we may introduce holes in the
|
|
|
|
* BA sequence space and/or make window accouting
|
|
|
|
* more difficult.
|
|
|
|
*
|
|
|
|
* XXX may want to control this with a driver
|
|
|
|
* capability; this may also change when we pull
|
|
|
|
* aggregation up into net80211
|
|
|
|
*/
|
2009-04-26 21:34:53 +00:00
|
|
|
seqno = ni->ni_txseqs[tid]++;
|
2008-09-21 23:00:19 +00:00
|
|
|
*(uint16_t *)wh->i_seq =
|
2009-04-26 21:34:53 +00:00
|
|
|
htole16(seqno << IEEE80211_SEQ_SEQ_SHIFT);
|
2009-04-27 17:39:41 +00:00
|
|
|
M_SEQNO_SET(m, seqno);
|
2008-09-21 23:00:19 +00:00
|
|
|
}
|
2004-12-08 17:26:47 +00:00
|
|
|
} else {
|
2009-04-26 21:34:53 +00:00
|
|
|
seqno = ni->ni_txseqs[IEEE80211_NONQOS_TID]++;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)wh->i_seq =
|
2009-04-26 21:34:53 +00:00
|
|
|
htole16(seqno << IEEE80211_SEQ_SEQ_SHIFT);
|
2009-04-27 17:39:41 +00:00
|
|
|
M_SEQNO_SET(m, seqno);
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
2009-04-27 17:39:41 +00:00
|
|
|
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
/* check if xmit fragmentation is required */
|
2008-04-20 20:35:46 +00:00
|
|
|
txfrag = (m->m_pkthdr.len > vap->iv_fragthreshold &&
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
!IEEE80211_IS_MULTICAST(wh->i_addr1) &&
|
2008-04-20 20:35:46 +00:00
|
|
|
(vap->iv_caps & IEEE80211_C_TXFRAG) &&
|
2009-04-26 21:37:02 +00:00
|
|
|
(m->m_flags & (M_FF | M_AMPDU_MPDU)) == 0);
|
2004-12-31 21:54:53 +00:00
|
|
|
if (key != NULL) {
|
|
|
|
/*
|
|
|
|
* IEEE 802.1X: send EAPOL frames always in the clear.
|
|
|
|
* WPA/WPA2: encrypt EAPOL keys when pairwise keys are set.
|
|
|
|
*/
|
2008-04-20 20:35:46 +00:00
|
|
|
if ((m->m_flags & M_EAPOL) == 0 ||
|
|
|
|
((vap->iv_flags & IEEE80211_F_WPA) &&
|
|
|
|
(vap->iv_opmode == IEEE80211_M_STA ?
|
2007-03-11 06:36:10 +00:00
|
|
|
!IEEE80211_KEY_UNDEFINED(key) :
|
|
|
|
!IEEE80211_KEY_UNDEFINED(&ni->ni_ucastkey)))) {
|
2004-12-31 21:54:53 +00:00
|
|
|
wh->i_fc[1] |= IEEE80211_FC1_WEP;
|
2008-04-20 20:35:46 +00:00
|
|
|
if (!ieee80211_crypto_enmic(vap, key, m, txfrag)) {
|
|
|
|
IEEE80211_NOTE_MAC(vap, IEEE80211_MSG_OUTPUT,
|
|
|
|
eh.ether_dhost,
|
|
|
|
"%s", "enmic failed, discard frame");
|
|
|
|
vap->iv_stats.is_crypto_enmicfail++;
|
2004-12-31 21:54:53 +00:00
|
|
|
goto bad;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
if (txfrag && !ieee80211_fragment(vap, m, hdrsize,
|
|
|
|
key != NULL ? key->wk_cipher->ic_header : 0, vap->iv_fragthreshold))
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
goto bad;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2008-10-26 00:55:38 +00:00
|
|
|
m->m_flags |= M_ENCAP; /* mark encapsulated */
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_data);
|
2009-04-26 21:34:53 +00:00
|
|
|
if (IEEE80211_IS_MULTICAST(wh->i_addr1)) {
|
2006-08-10 05:37:49 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_mcast);
|
2009-04-26 21:34:53 +00:00
|
|
|
m->m_flags |= M_MCAST;
|
|
|
|
} else
|
2006-08-10 05:37:49 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_ucast);
|
2004-12-08 17:26:47 +00:00
|
|
|
IEEE80211_NODE_STAT_ADD(ni, tx_bytes, datalen);
|
|
|
|
|
2003-06-23 16:55:01 +00:00
|
|
|
return m;
|
2003-08-19 22:17:04 +00:00
|
|
|
bad:
|
|
|
|
if (m != NULL)
|
|
|
|
m_freem(m);
|
|
|
|
return NULL;
|
2008-04-20 20:35:46 +00:00
|
|
|
#undef WH4
|
2012-05-01 15:56:26 +00:00
|
|
|
#undef MC01
|
2003-06-23 16:55:01 +00:00
|
|
|
}
|
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
/*
|
|
|
|
* Fragment the frame according to the specified mtu.
|
|
|
|
* The size of the 802.11 header (w/o padding) is provided
|
|
|
|
* so we don't need to recalculate it. We create a new
|
|
|
|
* mbuf for each fragment and chain it through m_nextpkt;
|
|
|
|
* we might be able to optimize this by reusing the original
|
|
|
|
* packet's mbufs but that is significantly more complicated.
|
|
|
|
*/
|
|
|
|
static int
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_fragment(struct ieee80211vap *vap, struct mbuf *m0,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
u_int hdrsize, u_int ciphdrsize, u_int mtu)
|
|
|
|
{
|
2013-05-25 06:28:30 +00:00
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
struct ieee80211_frame *wh, *whf;
|
|
|
|
struct mbuf *m, *prev, *next;
|
|
|
|
u_int totalhdrsize, fragno, fragsize, off, remainder, payload;
|
2013-05-25 06:28:30 +00:00
|
|
|
u_int hdrspace;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
|
|
|
|
KASSERT(m0->m_nextpkt == NULL, ("mbuf already chained?"));
|
|
|
|
KASSERT(m0->m_pkthdr.len > mtu,
|
|
|
|
("pktlen %u mtu %u", m0->m_pkthdr.len, mtu));
|
|
|
|
|
2013-05-25 06:28:30 +00:00
|
|
|
/*
|
|
|
|
* Honor driver DATAPAD requirement.
|
|
|
|
*/
|
|
|
|
if (ic->ic_flags & IEEE80211_F_DATAPAD)
|
|
|
|
hdrspace = roundup(hdrsize, sizeof(uint32_t));
|
|
|
|
else
|
|
|
|
hdrspace = hdrsize;
|
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
wh = mtod(m0, struct ieee80211_frame *);
|
|
|
|
/* NB: mark the first frag; it will be propagated below */
|
|
|
|
wh->i_fc[1] |= IEEE80211_FC1_MORE_FRAG;
|
2013-05-25 06:28:30 +00:00
|
|
|
totalhdrsize = hdrspace + ciphdrsize;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
fragno = 1;
|
|
|
|
off = mtu - ciphdrsize;
|
|
|
|
remainder = m0->m_pkthdr.len - off;
|
|
|
|
prev = m0;
|
|
|
|
do {
|
|
|
|
fragsize = totalhdrsize + remainder;
|
|
|
|
if (fragsize > mtu)
|
|
|
|
fragsize = mtu;
|
2008-04-20 20:35:46 +00:00
|
|
|
/* XXX fragsize can be >2048! */
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
KASSERT(fragsize < MCLBYTES,
|
|
|
|
("fragment size %u too big!", fragsize));
|
|
|
|
if (fragsize > MHLEN)
|
2012-12-05 08:04:20 +00:00
|
|
|
m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
else
|
2012-12-05 08:04:20 +00:00
|
|
|
m = m_gethdr(M_NOWAIT, MT_DATA);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
if (m == NULL)
|
|
|
|
goto bad;
|
|
|
|
/* leave room to prepend any cipher header */
|
|
|
|
m_align(m, fragsize - ciphdrsize);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Form the header in the fragment. Note that since
|
|
|
|
* we mark the first fragment with the MORE_FRAG bit
|
|
|
|
* it automatically is propagated to each fragment; we
|
|
|
|
* need only clear it on the last fragment (done below).
|
2012-03-04 05:52:26 +00:00
|
|
|
* NB: frag 1+ dont have Mesh Control field present.
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*/
|
|
|
|
whf = mtod(m, struct ieee80211_frame *);
|
|
|
|
memcpy(whf, wh, hdrsize);
|
2012-03-04 05:52:26 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS) {
|
|
|
|
if (IEEE80211_IS_DSTODS(wh))
|
|
|
|
((struct ieee80211_qosframe_addr4 *)
|
|
|
|
whf)->i_qos[1] &= ~IEEE80211_QOS_MC;
|
|
|
|
else
|
|
|
|
((struct ieee80211_qosframe *)
|
|
|
|
whf)->i_qos[1] &= ~IEEE80211_QOS_MC;
|
|
|
|
}
|
|
|
|
#endif
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)&whf->i_seq[0] |= htole16(
|
|
|
|
(fragno & IEEE80211_SEQ_FRAG_MASK) <<
|
|
|
|
IEEE80211_SEQ_FRAG_SHIFT);
|
|
|
|
fragno++;
|
|
|
|
|
|
|
|
payload = fragsize - totalhdrsize;
|
|
|
|
/* NB: destination is known to be contiguous */
|
2013-05-25 06:28:30 +00:00
|
|
|
|
|
|
|
m_copydata(m0, off, payload, mtod(m, uint8_t *) + hdrspace);
|
|
|
|
m->m_len = hdrspace + payload;
|
|
|
|
m->m_pkthdr.len = hdrspace + payload;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m->m_flags |= M_FRAG;
|
|
|
|
|
|
|
|
/* chain up the fragment */
|
|
|
|
prev->m_nextpkt = m;
|
|
|
|
prev = m;
|
|
|
|
|
|
|
|
/* deduct fragment just formed */
|
|
|
|
remainder -= payload;
|
|
|
|
off += payload;
|
|
|
|
} while (remainder != 0);
|
2009-02-09 04:39:16 +00:00
|
|
|
|
|
|
|
/* set the last fragment */
|
|
|
|
m->m_flags |= M_LASTFRAG;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
whf->i_fc[1] &= ~IEEE80211_FC1_MORE_FRAG;
|
|
|
|
|
|
|
|
/* strip first mbuf now that everything has been copied */
|
|
|
|
m_adj(m0, -(m0->m_pkthdr.len - (mtu - ciphdrsize)));
|
|
|
|
m0->m_flags |= M_FIRSTFRAG | M_FRAG;
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
vap->iv_stats.is_tx_fragframes++;
|
|
|
|
vap->iv_stats.is_tx_frags += fragno-1;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
|
|
|
|
return 1;
|
|
|
|
bad:
|
|
|
|
/* reclaim fragments but leave original frame for caller to free */
|
|
|
|
for (m = m0->m_nextpkt; m != NULL; m = next) {
|
|
|
|
next = m->m_nextpkt;
|
|
|
|
m->m_nextpkt = NULL; /* XXX paranoid */
|
|
|
|
m_freem(m);
|
|
|
|
}
|
|
|
|
m0->m_nextpkt = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2003-06-23 16:55:01 +00:00
|
|
|
/*
|
|
|
|
* Add a supported rates element id to a frame.
|
|
|
|
*/
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
uint8_t *
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ieee80211_add_rates(uint8_t *frm, const struct ieee80211_rateset *rs)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
|
|
|
int nrates;
|
|
|
|
|
|
|
|
*frm++ = IEEE80211_ELEMID_RATES;
|
|
|
|
nrates = rs->rs_nrates;
|
|
|
|
if (nrates > IEEE80211_RATE_SIZE)
|
|
|
|
nrates = IEEE80211_RATE_SIZE;
|
|
|
|
*frm++ = nrates;
|
|
|
|
memcpy(frm, rs->rs_rates, nrates);
|
|
|
|
return frm + nrates;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add an extended supported rates element id to a frame.
|
|
|
|
*/
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
uint8_t *
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ieee80211_add_xrates(uint8_t *frm, const struct ieee80211_rateset *rs)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Add an extended supported rates element if operating in 11g mode.
|
|
|
|
*/
|
|
|
|
if (rs->rs_nrates > IEEE80211_RATE_SIZE) {
|
|
|
|
int nrates = rs->rs_nrates - IEEE80211_RATE_SIZE;
|
|
|
|
*frm++ = IEEE80211_ELEMID_XRATES;
|
|
|
|
*frm++ = nrates;
|
|
|
|
memcpy(frm, rs->rs_rates + IEEE80211_RATE_SIZE, nrates);
|
|
|
|
frm += nrates;
|
|
|
|
}
|
|
|
|
return frm;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2008-04-20 20:35:46 +00:00
|
|
|
* Add an ssid element to a frame.
|
2003-06-23 16:55:01 +00:00
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_ssid(uint8_t *frm, const uint8_t *ssid, u_int len)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
|
|
|
*frm++ = IEEE80211_ELEMID_SSID;
|
|
|
|
*frm++ = len;
|
|
|
|
memcpy(frm, ssid, len);
|
|
|
|
return frm + len;
|
|
|
|
}
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
/*
|
|
|
|
* Add an erp element to a frame.
|
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_erp(uint8_t *frm, struct ieee80211com *ic)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
uint8_t erp;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
*frm++ = IEEE80211_ELEMID_ERP;
|
|
|
|
*frm++ = 1;
|
|
|
|
erp = 0;
|
|
|
|
if (ic->ic_nonerpsta != 0)
|
|
|
|
erp |= IEEE80211_ERP_NON_ERP_PRESENT;
|
|
|
|
if (ic->ic_flags & IEEE80211_F_USEPROT)
|
|
|
|
erp |= IEEE80211_ERP_USE_PROTECTION;
|
|
|
|
if (ic->ic_flags & IEEE80211_F_USEBARKER)
|
|
|
|
erp |= IEEE80211_ERP_LONG_PREAMBLE;
|
|
|
|
*frm++ = erp;
|
|
|
|
return frm;
|
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Add a CFParams element to a frame.
|
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static uint8_t *
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_add_cfparms(uint8_t *frm, struct ieee80211com *ic)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#define ADDSHORT(frm, v) do { \
|
|
|
|
LE_WRITE_2(frm, v); \
|
|
|
|
frm += 2; \
|
2004-12-08 17:26:47 +00:00
|
|
|
} while (0)
|
2008-04-20 20:35:46 +00:00
|
|
|
*frm++ = IEEE80211_ELEMID_CFPARMS;
|
|
|
|
*frm++ = 6;
|
|
|
|
*frm++ = 0; /* CFP count */
|
|
|
|
*frm++ = 2; /* CFP period */
|
|
|
|
ADDSHORT(frm, 0); /* CFP MaxDuration (TU) */
|
|
|
|
ADDSHORT(frm, 0); /* CFP CurRemaining (TU) */
|
2004-12-08 17:26:47 +00:00
|
|
|
return frm;
|
|
|
|
#undef ADDSHORT
|
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
static __inline uint8_t *
|
|
|
|
add_appie(uint8_t *frm, const struct ieee80211_appie *ie)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
memcpy(frm, ie->ie_data, ie->ie_len);
|
|
|
|
return frm + ie->ie_len;
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
static __inline uint8_t *
|
|
|
|
add_ie(uint8_t *frm, const uint8_t *ie)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
memcpy(frm, ie, 2 + ie[1]);
|
|
|
|
return frm + 2 + ie[1];
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
#define WME_OUI_BYTES 0x00, 0x50, 0xf2
|
|
|
|
/*
|
|
|
|
* Add a WME information element to a frame.
|
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_wme_info(uint8_t *frm, struct ieee80211_wme_state *wme)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
|
|
|
static const struct ieee80211_wme_info info = {
|
|
|
|
.wme_id = IEEE80211_ELEMID_VENDOR,
|
|
|
|
.wme_len = sizeof(struct ieee80211_wme_info) - 2,
|
|
|
|
.wme_oui = { WME_OUI_BYTES },
|
|
|
|
.wme_type = WME_OUI_TYPE,
|
|
|
|
.wme_subtype = WME_INFO_OUI_SUBTYPE,
|
|
|
|
.wme_version = WME_VERSION,
|
|
|
|
.wme_info = 0,
|
|
|
|
};
|
|
|
|
memcpy(frm, &info, sizeof(info));
|
|
|
|
return frm + sizeof(info);
|
2003-06-23 16:55:01 +00:00
|
|
|
}
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
/*
|
|
|
|
* Add a WME parameters element to a frame.
|
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_wme_param(uint8_t *frm, struct ieee80211_wme_state *wme)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
|
|
|
#define SM(_v, _f) (((_v) << _f##_S) & _f)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#define ADDSHORT(frm, v) do { \
|
|
|
|
LE_WRITE_2(frm, v); \
|
|
|
|
frm += 2; \
|
2004-12-08 17:26:47 +00:00
|
|
|
} while (0)
|
|
|
|
/* NB: this works 'cuz a param has an info at the front */
|
|
|
|
static const struct ieee80211_wme_info param = {
|
|
|
|
.wme_id = IEEE80211_ELEMID_VENDOR,
|
|
|
|
.wme_len = sizeof(struct ieee80211_wme_param) - 2,
|
|
|
|
.wme_oui = { WME_OUI_BYTES },
|
|
|
|
.wme_type = WME_OUI_TYPE,
|
|
|
|
.wme_subtype = WME_PARAM_OUI_SUBTYPE,
|
|
|
|
.wme_version = WME_VERSION,
|
|
|
|
};
|
|
|
|
int i;
|
|
|
|
|
|
|
|
memcpy(frm, ¶m, sizeof(param));
|
|
|
|
frm += __offsetof(struct ieee80211_wme_info, wme_info);
|
|
|
|
*frm++ = wme->wme_bssChanParams.cap_info; /* AC info */
|
|
|
|
*frm++ = 0; /* reserved field */
|
|
|
|
for (i = 0; i < WME_NUM_AC; i++) {
|
|
|
|
const struct wmeParams *ac =
|
|
|
|
&wme->wme_bssChanParams.cap_wmeParams[i];
|
|
|
|
*frm++ = SM(i, WME_PARAM_ACI)
|
|
|
|
| SM(ac->wmep_acm, WME_PARAM_ACM)
|
|
|
|
| SM(ac->wmep_aifsn, WME_PARAM_AIFSN)
|
|
|
|
;
|
|
|
|
*frm++ = SM(ac->wmep_logcwmax, WME_PARAM_LOGCWMAX)
|
|
|
|
| SM(ac->wmep_logcwmin, WME_PARAM_LOGCWMIN)
|
|
|
|
;
|
|
|
|
ADDSHORT(frm, ac->wmep_txopLimit);
|
|
|
|
}
|
|
|
|
return frm;
|
|
|
|
#undef SM
|
|
|
|
#undef ADDSHORT
|
|
|
|
}
|
|
|
|
#undef WME_OUI_BYTES
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Add an 11h Power Constraint element to a frame.
|
|
|
|
*/
|
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_powerconstraint(uint8_t *frm, struct ieee80211vap *vap)
|
|
|
|
{
|
|
|
|
const struct ieee80211_channel *c = vap->iv_bss->ni_chan;
|
|
|
|
/* XXX per-vap tx power limit? */
|
|
|
|
int8_t limit = vap->iv_ic->ic_txpowlimit / 2;
|
|
|
|
|
|
|
|
frm[0] = IEEE80211_ELEMID_PWRCNSTR;
|
|
|
|
frm[1] = 1;
|
|
|
|
frm[2] = c->ic_maxregpower > limit ? c->ic_maxregpower - limit : 0;
|
|
|
|
return frm + 3;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add an 11h Power Capability element to a frame.
|
|
|
|
*/
|
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_powercapability(uint8_t *frm, const struct ieee80211_channel *c)
|
|
|
|
{
|
|
|
|
frm[0] = IEEE80211_ELEMID_PWRCAP;
|
|
|
|
frm[1] = 2;
|
|
|
|
frm[2] = c->ic_minpower;
|
|
|
|
frm[3] = c->ic_maxpower;
|
|
|
|
return frm + 4;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add an 11h Supported Channels element to a frame.
|
|
|
|
*/
|
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_supportedchannels(uint8_t *frm, struct ieee80211com *ic)
|
|
|
|
{
|
|
|
|
static const int ielen = 26;
|
|
|
|
|
|
|
|
frm[0] = IEEE80211_ELEMID_SUPPCHAN;
|
|
|
|
frm[1] = ielen;
|
|
|
|
/* XXX not correct */
|
|
|
|
memcpy(frm+2, ic->ic_chan_avail, ielen);
|
|
|
|
return frm + 2 + ielen;
|
|
|
|
}
|
|
|
|
|
2011-11-08 04:00:24 +00:00
|
|
|
/*
|
|
|
|
* Add an 11h Quiet time element to a frame.
|
|
|
|
*/
|
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_quiet(uint8_t *frm, struct ieee80211vap *vap)
|
|
|
|
{
|
|
|
|
struct ieee80211_quiet_ie *quiet = (struct ieee80211_quiet_ie *) frm;
|
|
|
|
|
|
|
|
quiet->quiet_ie = IEEE80211_ELEMID_QUIET;
|
|
|
|
quiet->len = 6;
|
|
|
|
if (vap->iv_quiet_count_value == 1)
|
|
|
|
vap->iv_quiet_count_value = vap->iv_quiet_count;
|
|
|
|
else if (vap->iv_quiet_count_value > 1)
|
|
|
|
vap->iv_quiet_count_value--;
|
|
|
|
|
|
|
|
if (vap->iv_quiet_count_value == 0) {
|
|
|
|
/* value 0 is reserved as per 802.11h standerd */
|
|
|
|
vap->iv_quiet_count_value = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
quiet->tbttcount = vap->iv_quiet_count_value;
|
|
|
|
quiet->period = vap->iv_quiet_period;
|
|
|
|
quiet->duration = htole16(vap->iv_quiet_duration);
|
|
|
|
quiet->offset = htole16(vap->iv_quiet_offset);
|
|
|
|
return frm + sizeof(*quiet);
|
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Add an 11h Channel Switch Announcement element to a frame.
|
|
|
|
* Note that we use the per-vap CSA count to adjust the global
|
|
|
|
* counter so we can use this routine to form probe response
|
|
|
|
* frames and get the current count.
|
|
|
|
*/
|
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_csa(uint8_t *frm, struct ieee80211vap *vap)
|
|
|
|
{
|
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
|
|
|
struct ieee80211_csa_ie *csa = (struct ieee80211_csa_ie *) frm;
|
|
|
|
|
2009-06-04 15:57:38 +00:00
|
|
|
csa->csa_ie = IEEE80211_ELEMID_CSA;
|
2008-04-20 20:35:46 +00:00
|
|
|
csa->csa_len = 3;
|
|
|
|
csa->csa_mode = 1; /* XXX force quiet on channel */
|
|
|
|
csa->csa_newchan = ieee80211_chan2ieee(ic, ic->ic_csa_newchan);
|
|
|
|
csa->csa_count = ic->ic_csa_count - vap->iv_csa_count;
|
|
|
|
return frm + sizeof(*csa);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add an 11h country information element to a frame.
|
|
|
|
*/
|
|
|
|
static uint8_t *
|
|
|
|
ieee80211_add_countryie(uint8_t *frm, struct ieee80211com *ic)
|
|
|
|
{
|
|
|
|
|
|
|
|
if (ic->ic_countryie == NULL ||
|
|
|
|
ic->ic_countryie_chan != ic->ic_bsschan) {
|
|
|
|
/*
|
|
|
|
* Handle lazy construction of ie. This is done on
|
|
|
|
* first use and after a channel change that requires
|
|
|
|
* re-calculation.
|
|
|
|
*/
|
|
|
|
if (ic->ic_countryie != NULL)
|
|
|
|
free(ic->ic_countryie, M_80211_NODE_IE);
|
|
|
|
ic->ic_countryie = ieee80211_alloc_countryie(ic);
|
|
|
|
if (ic->ic_countryie == NULL)
|
|
|
|
return frm;
|
|
|
|
ic->ic_countryie_chan = ic->ic_bsschan;
|
|
|
|
}
|
|
|
|
return add_appie(frm, ic->ic_countryie);
|
|
|
|
}
|
|
|
|
|
2013-07-25 06:23:26 +00:00
|
|
|
uint8_t *
|
|
|
|
ieee80211_add_wpa(uint8_t *frm, const struct ieee80211vap *vap)
|
|
|
|
{
|
|
|
|
if (vap->iv_flags & IEEE80211_F_WPA1 && vap->iv_wpa_ie != NULL)
|
|
|
|
return (add_ie(frm, vap->iv_wpa_ie));
|
|
|
|
else {
|
|
|
|
/* XXX else complain? */
|
|
|
|
return (frm);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
uint8_t *
|
|
|
|
ieee80211_add_rsn(uint8_t *frm, const struct ieee80211vap *vap)
|
|
|
|
{
|
|
|
|
if (vap->iv_flags & IEEE80211_F_WPA2 && vap->iv_rsn_ie != NULL)
|
|
|
|
return (add_ie(frm, vap->iv_rsn_ie));
|
|
|
|
else {
|
|
|
|
/* XXX else complain? */
|
|
|
|
return (frm);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
uint8_t *
|
|
|
|
ieee80211_add_qos(uint8_t *frm, const struct ieee80211_node *ni)
|
|
|
|
{
|
|
|
|
if (ni->ni_flags & IEEE80211_NODE_QOS) {
|
|
|
|
*frm++ = IEEE80211_ELEMID_QOS;
|
|
|
|
*frm++ = 1;
|
|
|
|
*frm++ = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return (frm);
|
|
|
|
}
|
|
|
|
|
2005-07-22 21:11:26 +00:00
|
|
|
/*
|
|
|
|
* Send a probe request frame with the specified ssid
|
|
|
|
* and any optional information element data.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
ieee80211_send_probereq(struct ieee80211_node *ni,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
const uint8_t sa[IEEE80211_ADDR_LEN],
|
|
|
|
const uint8_t da[IEEE80211_ADDR_LEN],
|
|
|
|
const uint8_t bssid[IEEE80211_ADDR_LEN],
|
2008-04-20 20:35:46 +00:00
|
|
|
const uint8_t *ssid, size_t ssidlen)
|
2005-07-22 21:11:26 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
2005-07-22 21:11:26 +00:00
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
2008-10-26 00:48:20 +00:00
|
|
|
const struct ieee80211_txparam *tp;
|
|
|
|
struct ieee80211_bpf_params params;
|
2005-07-22 21:11:26 +00:00
|
|
|
struct ieee80211_frame *wh;
|
2006-12-27 18:46:18 +00:00
|
|
|
const struct ieee80211_rateset *rs;
|
2005-07-22 21:11:26 +00:00
|
|
|
struct mbuf *m;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
uint8_t *frm;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
int ret;
|
2005-07-22 21:11:26 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_state == IEEE80211_S_CAC) {
|
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_OUTPUT, ni,
|
|
|
|
"block %s frame in CAC state", "probe request");
|
|
|
|
vap->iv_stats.is_tx_badstate++;
|
|
|
|
return EIO; /* XXX */
|
|
|
|
}
|
|
|
|
|
2005-07-22 21:11:26 +00:00
|
|
|
/*
|
|
|
|
* Hold a reference on the node so it doesn't go away until after
|
|
|
|
* the xmit is complete all the way in the driver. On error we
|
|
|
|
* will remove our reference.
|
|
|
|
*/
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_NODE,
|
2005-07-22 21:11:26 +00:00
|
|
|
"ieee80211_ref_node (%s:%u) %p<%s> refcnt %d\n",
|
|
|
|
__func__, __LINE__,
|
|
|
|
ni, ether_sprintf(ni->ni_macaddr),
|
|
|
|
ieee80211_node_refcnt(ni)+1);
|
|
|
|
ieee80211_ref_node(ni);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* prreq frame format
|
|
|
|
* [tlv] ssid
|
|
|
|
* [tlv] supported rates
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] RSN (optional)
|
2005-07-22 21:11:26 +00:00
|
|
|
* [tlv] extended supported rates
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] WPA (optional)
|
2005-07-22 21:11:26 +00:00
|
|
|
* [tlv] user-specified ie's
|
|
|
|
*/
|
|
|
|
m = ieee80211_getmgtframe(&frm,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
2008-04-20 20:35:46 +00:00
|
|
|
2 + IEEE80211_NWID_LEN
|
2005-07-22 21:11:26 +00:00
|
|
|
+ 2 + IEEE80211_RATE_SIZE
|
2008-04-20 20:35:46 +00:00
|
|
|
+ sizeof(struct ieee80211_ie_wpa)
|
2005-07-22 21:11:26 +00:00
|
|
|
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE)
|
2008-04-20 20:35:46 +00:00
|
|
|
+ sizeof(struct ieee80211_ie_wpa)
|
|
|
|
+ (vap->iv_appie_probereq != NULL ?
|
|
|
|
vap->iv_appie_probereq->ie_len : 0)
|
2005-07-22 21:11:26 +00:00
|
|
|
);
|
|
|
|
if (m == NULL) {
|
2008-04-20 20:35:46 +00:00
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
2005-07-22 21:11:26 +00:00
|
|
|
ieee80211_free_node(ni);
|
|
|
|
return ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
frm = ieee80211_add_ssid(frm, ssid, ssidlen);
|
2006-12-27 18:46:18 +00:00
|
|
|
rs = ieee80211_get_suprates(ic, ic->ic_curchan);
|
|
|
|
frm = ieee80211_add_rates(frm, rs);
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_rsn(frm, vap);
|
2006-12-27 18:46:18 +00:00
|
|
|
frm = ieee80211_add_xrates(frm, rs);
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_wpa(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_appie_probereq != NULL)
|
|
|
|
frm = add_appie(frm, vap->iv_appie_probereq);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = frm - mtod(m, uint8_t *);
|
2005-07-22 21:11:26 +00:00
|
|
|
|
2008-10-26 00:48:20 +00:00
|
|
|
KASSERT(M_LEADINGSPACE(m) >= sizeof(struct ieee80211_frame),
|
|
|
|
("leading space %zd", M_LEADINGSPACE(m)));
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, sizeof(struct ieee80211_frame), M_NOWAIT);
|
2008-10-26 00:48:20 +00:00
|
|
|
if (m == NULL) {
|
|
|
|
/* NB: cannot happen */
|
|
|
|
ieee80211_free_node(ni);
|
2005-07-22 21:11:26 +00:00
|
|
|
return ENOMEM;
|
2008-10-26 00:48:20 +00:00
|
|
|
}
|
2005-07-22 21:11:26 +00:00
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
IEEE80211_TX_LOCK(ic);
|
2005-07-22 21:11:26 +00:00
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
2009-04-26 21:34:53 +00:00
|
|
|
ieee80211_send_setup(ni, m,
|
2008-10-26 00:43:11 +00:00
|
|
|
IEEE80211_FC0_TYPE_MGT | IEEE80211_FC0_SUBTYPE_PROBE_REQ,
|
|
|
|
IEEE80211_NONQOS_TID, sa, da, bssid);
|
2005-07-22 21:11:26 +00:00
|
|
|
/* XXX power management? */
|
2008-10-26 00:55:38 +00:00
|
|
|
m->m_flags |= M_ENCAP; /* mark encapsulated */
|
2005-07-22 21:11:26 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
M_WME_SETAC(m, WME_AC_BE);
|
|
|
|
|
2005-07-22 21:11:26 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_probereq);
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_mgmt);
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_DEBUG | IEEE80211_MSG_DUMPPKTS,
|
|
|
|
"send probe req on channel %u bssid %s ssid \"%.*s\"\n",
|
|
|
|
ieee80211_chan2ieee(ic, ic->ic_curchan), ether_sprintf(bssid),
|
|
|
|
ssidlen, ssid);
|
2005-07-22 21:11:26 +00:00
|
|
|
|
2008-10-26 00:48:20 +00:00
|
|
|
memset(¶ms, 0, sizeof(params));
|
|
|
|
params.ibp_pri = M_WME_GETAC(m);
|
|
|
|
tp = &vap->iv_txparms[ieee80211_chan2mode(ic->ic_curchan)];
|
|
|
|
params.ibp_rate0 = tp->mgmtrate;
|
|
|
|
if (IEEE80211_IS_MULTICAST(da)) {
|
|
|
|
params.ibp_flags |= IEEE80211_BPF_NOACK;
|
|
|
|
params.ibp_try0 = 1;
|
|
|
|
} else
|
|
|
|
params.ibp_try0 = tp->maxretry;
|
|
|
|
params.ibp_power = ni->ni_txpower;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
ret = ieee80211_raw_output(vap, ni, m, ¶ms);
|
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
return (ret);
|
2005-07-22 21:11:26 +00:00
|
|
|
}
|
|
|
|
|
2006-02-25 05:36:43 +00:00
|
|
|
/*
|
|
|
|
* Calculate capability information for mgt frames.
|
|
|
|
*/
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
uint16_t
|
|
|
|
ieee80211_getcapinfo(struct ieee80211vap *vap, struct ieee80211_channel *chan)
|
2006-02-25 05:36:43 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
uint16_t capinfo;
|
2006-02-25 05:36:43 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
KASSERT(vap->iv_opmode != IEEE80211_M_STA, ("station mode"));
|
2006-02-25 05:36:43 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_opmode == IEEE80211_M_HOSTAP)
|
2006-02-25 05:36:43 +00:00
|
|
|
capinfo = IEEE80211_CAPINFO_ESS;
|
2008-04-20 20:35:46 +00:00
|
|
|
else if (vap->iv_opmode == IEEE80211_M_IBSS)
|
2006-02-25 05:36:43 +00:00
|
|
|
capinfo = IEEE80211_CAPINFO_IBSS;
|
|
|
|
else
|
|
|
|
capinfo = 0;
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_PRIVACY)
|
2006-02-25 05:36:43 +00:00
|
|
|
capinfo |= IEEE80211_CAPINFO_PRIVACY;
|
|
|
|
if ((ic->ic_flags & IEEE80211_F_SHPREAMBLE) &&
|
|
|
|
IEEE80211_IS_CHAN_2GHZ(chan))
|
|
|
|
capinfo |= IEEE80211_CAPINFO_SHORT_PREAMBLE;
|
|
|
|
if (ic->ic_flags & IEEE80211_F_SHSLOT)
|
|
|
|
capinfo |= IEEE80211_CAPINFO_SHORT_SLOTTIME;
|
2008-04-20 20:35:46 +00:00
|
|
|
if (IEEE80211_IS_CHAN_5GHZ(chan) && (vap->iv_flags & IEEE80211_F_DOTH))
|
|
|
|
capinfo |= IEEE80211_CAPINFO_SPECTRUM_MGMT;
|
2006-02-25 05:36:43 +00:00
|
|
|
return capinfo;
|
|
|
|
}
|
|
|
|
|
2003-08-19 22:17:04 +00:00
|
|
|
/*
|
|
|
|
* Send a management frame. The node is for the destination (or ic_bss
|
|
|
|
* when in station mode). Nodes other than ic_bss have their reference
|
|
|
|
* count bumped to reflect our use for an indeterminant time.
|
|
|
|
*/
|
2003-06-23 16:55:01 +00:00
|
|
|
int
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_send_mgmt(struct ieee80211_node *ni, int type, int arg)
|
2003-06-23 16:55:01 +00:00
|
|
|
{
|
2007-11-02 05:22:25 +00:00
|
|
|
#define HTFLAGS (IEEE80211_NODE_HT | IEEE80211_NODE_HTCOMPAT)
|
2008-04-20 20:35:46 +00:00
|
|
|
#define senderr(_x, _v) do { vap->iv_stats._v++; ret = _x; goto bad; } while (0)
|
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
|
|
|
struct ieee80211_node *bss = vap->iv_bss;
|
2008-10-26 00:43:11 +00:00
|
|
|
struct ieee80211_bpf_params params;
|
2003-06-23 16:55:01 +00:00
|
|
|
struct mbuf *m;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
uint8_t *frm;
|
|
|
|
uint16_t capinfo;
|
|
|
|
int has_challenge, is_shared_key, ret, status;
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2003-08-19 22:17:04 +00:00
|
|
|
KASSERT(ni != NULL, ("null node"));
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Hold a reference on the node so it doesn't go away until after
|
|
|
|
* the xmit is complete all the way in the driver. On error we
|
2008-04-20 20:35:46 +00:00
|
|
|
* will remove our reference.
|
|
|
|
*/
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_NODE,
|
|
|
|
"ieee80211_ref_node (%s:%u) %p<%s> refcnt %d\n",
|
|
|
|
__func__, __LINE__,
|
|
|
|
ni, ether_sprintf(ni->ni_macaddr),
|
|
|
|
ieee80211_node_refcnt(ni)+1);
|
|
|
|
ieee80211_ref_node(ni);
|
2003-10-17 22:09:20 +00:00
|
|
|
|
2008-10-26 00:43:11 +00:00
|
|
|
memset(¶ms, 0, sizeof(params));
|
2008-04-20 20:35:46 +00:00
|
|
|
switch (type) {
|
2003-06-23 16:55:01 +00:00
|
|
|
|
|
|
|
case IEEE80211_FC0_SUBTYPE_AUTH:
|
2004-12-08 17:26:47 +00:00
|
|
|
status = arg >> 16;
|
|
|
|
arg &= 0xffff;
|
|
|
|
has_challenge = ((arg == IEEE80211_AUTH_SHARED_CHALLENGE ||
|
|
|
|
arg == IEEE80211_AUTH_SHARED_RESPONSE) &&
|
|
|
|
ni->ni_challenge != NULL);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Deduce whether we're doing open authentication or
|
|
|
|
* shared key authentication. We do the latter if
|
|
|
|
* we're in the middle of a shared key authentication
|
|
|
|
* handshake or if we're initiating an authentication
|
|
|
|
* request and configured to use shared key.
|
|
|
|
*/
|
|
|
|
is_shared_key = has_challenge ||
|
|
|
|
arg >= IEEE80211_AUTH_SHARED_RESPONSE ||
|
|
|
|
(arg == IEEE80211_AUTH_SHARED_REQUEST &&
|
2008-04-20 20:35:46 +00:00
|
|
|
bss->ni_authmode == IEEE80211_AUTH_SHARED);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
m = ieee80211_getmgtframe(&frm,
|
2008-10-26 00:43:11 +00:00
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
3 * sizeof(uint16_t)
|
2004-12-08 17:26:47 +00:00
|
|
|
+ (has_challenge && status == IEEE80211_STATUS_SUCCESS ?
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
sizeof(uint16_t)+IEEE80211_CHALLENGE_LEN : 0)
|
2004-12-08 17:26:47 +00:00
|
|
|
);
|
2003-06-23 16:55:01 +00:00
|
|
|
if (m == NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
senderr(ENOMEM, is_tx_nobuf);
|
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
((uint16_t *)frm)[0] =
|
2004-12-08 17:26:47 +00:00
|
|
|
(is_shared_key) ? htole16(IEEE80211_AUTH_ALG_SHARED)
|
|
|
|
: htole16(IEEE80211_AUTH_ALG_OPEN);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
((uint16_t *)frm)[1] = htole16(arg); /* sequence number */
|
|
|
|
((uint16_t *)frm)[2] = htole16(status);/* status */
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
if (has_challenge && status == IEEE80211_STATUS_SUCCESS) {
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
((uint16_t *)frm)[3] =
|
2004-12-08 17:26:47 +00:00
|
|
|
htole16((IEEE80211_CHALLENGE_LEN << 8) |
|
|
|
|
IEEE80211_ELEMID_CHALLENGE);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
memcpy(&((uint16_t *)frm)[4], ni->ni_challenge,
|
2004-12-08 17:26:47 +00:00
|
|
|
IEEE80211_CHALLENGE_LEN);
|
|
|
|
m->m_pkthdr.len = m->m_len =
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
4 * sizeof(uint16_t) + IEEE80211_CHALLENGE_LEN;
|
2004-12-08 17:26:47 +00:00
|
|
|
if (arg == IEEE80211_AUTH_SHARED_RESPONSE) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_AUTH, ni,
|
|
|
|
"request encrypt frame (%s)", __func__);
|
2008-10-26 00:43:11 +00:00
|
|
|
/* mark frame for encryption */
|
|
|
|
params.ibp_flags |= IEEE80211_BPF_CRYPTO;
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
} else
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = 3 * sizeof(uint16_t);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
/* XXX not right for shared key */
|
|
|
|
if (status == IEEE80211_STATUS_SUCCESS)
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_auth);
|
|
|
|
else
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_auth_fail);
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_opmode == IEEE80211_M_STA)
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ieee80211_add_callback(m, ieee80211_tx_mgt_cb,
|
2008-04-20 20:35:46 +00:00
|
|
|
(void *) vap->iv_state);
|
2003-06-23 16:55:01 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case IEEE80211_FC0_SUBTYPE_DEAUTH:
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_AUTH, ni,
|
|
|
|
"send station deauthenticate (reason %d)", arg);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m = ieee80211_getmgtframe(&frm,
|
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
|
|
|
sizeof(uint16_t));
|
2003-06-23 16:55:01 +00:00
|
|
|
if (m == NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
senderr(ENOMEM, is_tx_nobuf);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(arg); /* reason */
|
|
|
|
m->m_pkthdr.len = m->m_len = sizeof(uint16_t);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_deauth);
|
|
|
|
IEEE80211_NODE_STAT_SET(ni, tx_deauth_code, arg);
|
|
|
|
|
2005-07-22 17:36:12 +00:00
|
|
|
ieee80211_node_unauthorize(ni); /* port closed */
|
2003-06-23 16:55:01 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case IEEE80211_FC0_SUBTYPE_ASSOC_REQ:
|
|
|
|
case IEEE80211_FC0_SUBTYPE_REASSOC_REQ:
|
|
|
|
/*
|
|
|
|
* asreq frame format
|
|
|
|
* [2] capability information
|
|
|
|
* [2] listen interval
|
|
|
|
* [6*] current AP address (reassoc only)
|
|
|
|
* [tlv] ssid
|
|
|
|
* [tlv] supported rates
|
|
|
|
* [tlv] extended supported rates
|
2008-04-20 20:35:46 +00:00
|
|
|
* [4] power capability (optional)
|
|
|
|
* [28] supported channels (optional)
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
* [tlv] HT capabilities
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] WME (optional)
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
* [tlv] Vendor OUI HT capabilities (optional)
|
|
|
|
* [tlv] Atheros capabilities (if negotiated)
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] AppIE's (optional)
|
2003-06-23 16:55:01 +00:00
|
|
|
*/
|
2004-12-08 17:26:47 +00:00
|
|
|
m = ieee80211_getmgtframe(&frm,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
|
|
|
sizeof(uint16_t)
|
|
|
|
+ sizeof(uint16_t)
|
2003-06-23 16:55:01 +00:00
|
|
|
+ IEEE80211_ADDR_LEN
|
2004-12-08 17:26:47 +00:00
|
|
|
+ 2 + IEEE80211_NWID_LEN
|
2003-06-23 16:55:01 +00:00
|
|
|
+ 2 + IEEE80211_RATE_SIZE
|
2004-12-08 17:26:47 +00:00
|
|
|
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE)
|
2008-04-20 20:35:46 +00:00
|
|
|
+ 4
|
|
|
|
+ 2 + 26
|
2004-12-08 17:26:47 +00:00
|
|
|
+ sizeof(struct ieee80211_wme_info)
|
2008-04-20 20:35:46 +00:00
|
|
|
+ sizeof(struct ieee80211_ie_htcap)
|
|
|
|
+ 4 + sizeof(struct ieee80211_ie_htcap)
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
+ sizeof(struct ieee80211_ath_ie)
|
2009-03-24 20:39:08 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
+ (vap->iv_appie_wpa != NULL ?
|
|
|
|
vap->iv_appie_wpa->ie_len : 0)
|
|
|
|
+ (vap->iv_appie_assocreq != NULL ?
|
|
|
|
vap->iv_appie_assocreq->ie_len : 0)
|
2004-12-08 17:26:47 +00:00
|
|
|
);
|
2003-06-23 16:55:01 +00:00
|
|
|
if (m == NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
senderr(ENOMEM, is_tx_nobuf);
|
2003-06-23 16:55:01 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
KASSERT(vap->iv_opmode == IEEE80211_M_STA,
|
|
|
|
("wrong mode %u", vap->iv_opmode));
|
2006-02-25 05:36:43 +00:00
|
|
|
capinfo = IEEE80211_CAPINFO_ESS;
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_PRIVACY)
|
2003-06-23 16:55:01 +00:00
|
|
|
capinfo |= IEEE80211_CAPINFO_PRIVACY;
|
2003-09-14 22:34:24 +00:00
|
|
|
/*
|
|
|
|
* NB: Some 11a AP's reject the request when
|
|
|
|
* short premable is set.
|
|
|
|
*/
|
|
|
|
if ((ic->ic_flags & IEEE80211_F_SHPREAMBLE) &&
|
2005-08-10 16:22:30 +00:00
|
|
|
IEEE80211_IS_CHAN_2GHZ(ic->ic_curchan))
|
2003-06-23 16:55:01 +00:00
|
|
|
capinfo |= IEEE80211_CAPINFO_SHORT_PREAMBLE;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
if (IEEE80211_IS_CHAN_ANYG(ic->ic_curchan) &&
|
2004-12-08 17:26:47 +00:00
|
|
|
(ic->ic_caps & IEEE80211_C_SHSLOT))
|
2003-06-23 16:55:01 +00:00
|
|
|
capinfo |= IEEE80211_CAPINFO_SHORT_SLOTTIME;
|
2007-11-02 05:22:25 +00:00
|
|
|
if ((ni->ni_capinfo & IEEE80211_CAPINFO_SPECTRUM_MGMT) &&
|
2008-04-20 20:35:46 +00:00
|
|
|
(vap->iv_flags & IEEE80211_F_DOTH))
|
2007-11-02 05:22:25 +00:00
|
|
|
capinfo |= IEEE80211_CAPINFO_SPECTRUM_MGMT;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(capinfo);
|
2003-06-23 16:55:01 +00:00
|
|
|
frm += 2;
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
KASSERT(bss->ni_intval != 0, ("beacon interval is zero!"));
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(howmany(ic->ic_lintval,
|
2008-04-20 20:35:46 +00:00
|
|
|
bss->ni_intval));
|
2003-06-23 16:55:01 +00:00
|
|
|
frm += 2;
|
|
|
|
|
|
|
|
if (type == IEEE80211_FC0_SUBTYPE_REASSOC_REQ) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_ADDR_COPY(frm, bss->ni_bssid);
|
2003-06-23 16:55:01 +00:00
|
|
|
frm += IEEE80211_ADDR_LEN;
|
|
|
|
}
|
|
|
|
|
|
|
|
frm = ieee80211_add_ssid(frm, ni->ni_essid, ni->ni_esslen);
|
|
|
|
frm = ieee80211_add_rates(frm, &ni->ni_rates);
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_rsn(frm, vap);
|
2003-06-23 16:55:01 +00:00
|
|
|
frm = ieee80211_add_xrates(frm, &ni->ni_rates);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (capinfo & IEEE80211_CAPINFO_SPECTRUM_MGMT) {
|
|
|
|
frm = ieee80211_add_powercapability(frm,
|
|
|
|
ic->ic_curchan);
|
|
|
|
frm = ieee80211_add_supportedchannels(frm, ic);
|
|
|
|
}
|
2009-06-07 22:00:22 +00:00
|
|
|
if ((vap->iv_flags_ht & IEEE80211_FHT_HT) &&
|
2008-04-20 20:35:46 +00:00
|
|
|
ni->ni_ies.htcap_ie != NULL &&
|
|
|
|
ni->ni_ies.htcap_ie[0] == IEEE80211_ELEMID_HTCAP)
|
2007-11-02 05:22:25 +00:00
|
|
|
frm = ieee80211_add_htcap(frm, ni);
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_wpa(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
if ((ic->ic_flags & IEEE80211_F_WME) &&
|
|
|
|
ni->ni_ies.wme_ie != NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
frm = ieee80211_add_wme_info(frm, &ic->ic_wme);
|
2009-06-07 22:00:22 +00:00
|
|
|
if ((vap->iv_flags_ht & IEEE80211_FHT_HT) &&
|
2008-04-20 20:35:46 +00:00
|
|
|
ni->ni_ies.htcap_ie != NULL &&
|
|
|
|
ni->ni_ies.htcap_ie[0] == IEEE80211_ELEMID_VENDOR)
|
2007-11-02 05:22:25 +00:00
|
|
|
frm = ieee80211_add_htcap_vendor(frm, ni);
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2009-03-26 19:13:11 +00:00
|
|
|
if (IEEE80211_ATH_CAP(vap, ni, IEEE80211_F_ATHEROS)) {
|
|
|
|
frm = ieee80211_add_ath(frm,
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_ATH_CAP(vap, ni, IEEE80211_F_ATHEROS),
|
2009-03-26 19:13:11 +00:00
|
|
|
((vap->iv_flags & IEEE80211_F_WPA) == 0 &&
|
|
|
|
ni->ni_authmode != IEEE80211_AUTH_8021X) ?
|
|
|
|
vap->iv_def_txkey : IEEE80211_KEYIX_NONE);
|
|
|
|
}
|
2009-03-24 20:39:08 +00:00
|
|
|
#endif /* IEEE80211_SUPPORT_SUPERG */
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_appie_assocreq != NULL)
|
|
|
|
frm = add_appie(frm, vap->iv_appie_assocreq);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = frm - mtod(m, uint8_t *);
|
2003-06-23 16:55:01 +00:00
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ieee80211_add_callback(m, ieee80211_tx_mgt_cb,
|
2008-04-20 20:35:46 +00:00
|
|
|
(void *) vap->iv_state);
|
2003-06-23 16:55:01 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case IEEE80211_FC0_SUBTYPE_ASSOC_RESP:
|
|
|
|
case IEEE80211_FC0_SUBTYPE_REASSOC_RESP:
|
|
|
|
/*
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
* asresp frame format
|
2003-06-23 16:55:01 +00:00
|
|
|
* [2] capability information
|
|
|
|
* [2] status
|
|
|
|
* [2] association ID
|
|
|
|
* [tlv] supported rates
|
|
|
|
* [tlv] extended supported rates
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] HT capabilities (standard, if STA enabled)
|
|
|
|
* [tlv] HT information (standard, if STA enabled)
|
|
|
|
* [tlv] WME (if configured and STA enabled)
|
|
|
|
* [tlv] HT capabilities (vendor OUI, if STA enabled)
|
|
|
|
* [tlv] HT information (vendor OUI, if STA enabled)
|
|
|
|
* [tlv] Atheros capabilities (if STA enabled)
|
|
|
|
* [tlv] AppIE's (optional)
|
2003-06-23 16:55:01 +00:00
|
|
|
*/
|
2004-12-08 17:26:47 +00:00
|
|
|
m = ieee80211_getmgtframe(&frm,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
|
|
|
sizeof(uint16_t)
|
|
|
|
+ sizeof(uint16_t)
|
|
|
|
+ sizeof(uint16_t)
|
2003-06-23 16:55:01 +00:00
|
|
|
+ 2 + IEEE80211_RATE_SIZE
|
2004-12-08 17:26:47 +00:00
|
|
|
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE)
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
+ sizeof(struct ieee80211_ie_htcap) + 4
|
|
|
|
+ sizeof(struct ieee80211_ie_htinfo) + 4
|
2008-04-20 20:35:46 +00:00
|
|
|
+ sizeof(struct ieee80211_wme_param)
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
+ sizeof(struct ieee80211_ath_ie)
|
2009-03-24 20:39:08 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
+ (vap->iv_appie_assocresp != NULL ?
|
|
|
|
vap->iv_appie_assocresp->ie_len : 0)
|
2004-12-08 17:26:47 +00:00
|
|
|
);
|
2003-06-23 16:55:01 +00:00
|
|
|
if (m == NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
senderr(ENOMEM, is_tx_nobuf);
|
2003-06-23 16:55:01 +00:00
|
|
|
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
capinfo = ieee80211_getcapinfo(vap, bss->ni_chan);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(capinfo);
|
2003-06-23 16:55:01 +00:00
|
|
|
frm += 2;
|
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(arg); /* status */
|
2003-06-23 16:55:01 +00:00
|
|
|
frm += 2;
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
if (arg == IEEE80211_STATUS_SUCCESS) {
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(ni->ni_associd);
|
2004-12-08 17:26:47 +00:00
|
|
|
IEEE80211_NODE_STAT(ni, tx_assoc);
|
|
|
|
} else
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_assoc_fail);
|
2003-06-23 16:55:01 +00:00
|
|
|
frm += 2;
|
|
|
|
|
2003-08-19 22:17:04 +00:00
|
|
|
frm = ieee80211_add_rates(frm, &ni->ni_rates);
|
|
|
|
frm = ieee80211_add_xrates(frm, &ni->ni_rates);
|
2007-11-02 05:22:25 +00:00
|
|
|
/* NB: respond according to what we received */
|
|
|
|
if ((ni->ni_flags & HTFLAGS) == IEEE80211_NODE_HT) {
|
|
|
|
frm = ieee80211_add_htcap(frm, ni);
|
|
|
|
frm = ieee80211_add_htinfo(frm, ni);
|
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
if ((vap->iv_flags & IEEE80211_F_WME) &&
|
|
|
|
ni->ni_ies.wme_ie != NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
frm = ieee80211_add_wme_param(frm, &ic->ic_wme);
|
2007-11-02 05:22:25 +00:00
|
|
|
if ((ni->ni_flags & HTFLAGS) == HTFLAGS) {
|
|
|
|
frm = ieee80211_add_htcap_vendor(frm, ni);
|
|
|
|
frm = ieee80211_add_htinfo_vendor(frm, ni);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
}
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2008-04-20 20:35:46 +00:00
|
|
|
if (IEEE80211_ATH_CAP(vap, ni, IEEE80211_F_ATHEROS))
|
2009-03-26 19:13:11 +00:00
|
|
|
frm = ieee80211_add_ath(frm,
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_ATH_CAP(vap, ni, IEEE80211_F_ATHEROS),
|
2009-03-26 19:13:11 +00:00
|
|
|
((vap->iv_flags & IEEE80211_F_WPA) == 0 &&
|
|
|
|
ni->ni_authmode != IEEE80211_AUTH_8021X) ?
|
|
|
|
vap->iv_def_txkey : IEEE80211_KEYIX_NONE);
|
2009-03-24 20:39:08 +00:00
|
|
|
#endif /* IEEE80211_SUPPORT_SUPERG */
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_appie_assocresp != NULL)
|
|
|
|
frm = add_appie(frm, vap->iv_appie_assocresp);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = frm - mtod(m, uint8_t *);
|
2003-06-23 16:55:01 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
case IEEE80211_FC0_SUBTYPE_DISASSOC:
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_ASSOC, ni,
|
|
|
|
"send station disassociate (reason %d)", arg);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m = ieee80211_getmgtframe(&frm,
|
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
|
|
|
sizeof(uint16_t));
|
2003-06-23 16:55:01 +00:00
|
|
|
if (m == NULL)
|
2004-12-08 17:26:47 +00:00
|
|
|
senderr(ENOMEM, is_tx_nobuf);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(arg); /* reason */
|
|
|
|
m->m_pkthdr.len = m->m_len = sizeof(uint16_t);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
IEEE80211_NODE_STAT(ni, tx_disassoc);
|
|
|
|
IEEE80211_NODE_STAT_SET(ni, tx_disassoc_code, arg);
|
2003-06-23 16:55:01 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
default:
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_ANY, ni,
|
|
|
|
"invalid mgmt frame type %u", type);
|
2003-10-17 23:15:30 +00:00
|
|
|
senderr(EINVAL, is_tx_unknownmgt);
|
2003-08-19 22:17:04 +00:00
|
|
|
/* NOTREACHED */
|
2003-06-23 16:55:01 +00:00
|
|
|
}
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
|
2008-10-26 00:43:11 +00:00
|
|
|
/* NB: force non-ProbeResp frames to the highest queue */
|
|
|
|
params.ibp_pri = WME_AC_VO;
|
|
|
|
params.ibp_rate0 = bss->ni_txparms->mgmtrate;
|
|
|
|
/* NB: we know all frames are unicast */
|
|
|
|
params.ibp_try0 = bss->ni_txparms->maxretry;
|
|
|
|
params.ibp_power = bss->ni_txpower;
|
|
|
|
return ieee80211_mgmt_output(ni, m, type, ¶ms);
|
2003-08-19 22:17:04 +00:00
|
|
|
bad:
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
ieee80211_free_node(ni);
|
2003-06-23 16:55:01 +00:00
|
|
|
return ret;
|
2003-08-19 22:17:04 +00:00
|
|
|
#undef senderr
|
2007-11-02 05:22:25 +00:00
|
|
|
#undef HTFLAGS
|
2003-06-23 16:55:01 +00:00
|
|
|
}
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* Return an mbuf with a probe response frame in it.
|
|
|
|
* Space is left to prepend and 802.11 header at the
|
|
|
|
* front but it's left to the caller to fill in.
|
|
|
|
*/
|
|
|
|
struct mbuf *
|
|
|
|
ieee80211_alloc_proberesp(struct ieee80211_node *bss, int legacy)
|
|
|
|
{
|
|
|
|
struct ieee80211vap *vap = bss->ni_vap;
|
|
|
|
struct ieee80211com *ic = bss->ni_ic;
|
|
|
|
const struct ieee80211_rateset *rs;
|
|
|
|
struct mbuf *m;
|
|
|
|
uint16_t capinfo;
|
|
|
|
uint8_t *frm;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* probe response frame format
|
|
|
|
* [8] time stamp
|
|
|
|
* [2] beacon interval
|
|
|
|
* [2] cabability information
|
|
|
|
* [tlv] ssid
|
|
|
|
* [tlv] supported rates
|
|
|
|
* [tlv] parameter set (FH/DS)
|
|
|
|
* [tlv] parameter set (IBSS)
|
|
|
|
* [tlv] country (optional)
|
|
|
|
* [3] power control (optional)
|
|
|
|
* [5] channel switch announcement (CSA) (optional)
|
|
|
|
* [tlv] extended rate phy (ERP)
|
|
|
|
* [tlv] extended supported rates
|
2008-07-07 17:24:55 +00:00
|
|
|
* [tlv] RSN (optional)
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] HT capabilities
|
|
|
|
* [tlv] HT information
|
|
|
|
* [tlv] WPA (optional)
|
|
|
|
* [tlv] WME (optional)
|
|
|
|
* [tlv] Vendor OUI HT capabilities (optional)
|
|
|
|
* [tlv] Vendor OUI HT information (optional)
|
|
|
|
* [tlv] Atheros capabilities
|
|
|
|
* [tlv] AppIE's (optional)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
* [tlv] Mesh ID (MBSS)
|
|
|
|
* [tlv] Mesh Conf (MBSS)
|
2008-04-20 20:35:46 +00:00
|
|
|
*/
|
|
|
|
m = ieee80211_getmgtframe(&frm,
|
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame),
|
|
|
|
8
|
|
|
|
+ sizeof(uint16_t)
|
|
|
|
+ sizeof(uint16_t)
|
|
|
|
+ 2 + IEEE80211_NWID_LEN
|
|
|
|
+ 2 + IEEE80211_RATE_SIZE
|
|
|
|
+ 7 /* max(7,3) */
|
|
|
|
+ IEEE80211_COUNTRY_MAX_SIZE
|
|
|
|
+ 3
|
|
|
|
+ sizeof(struct ieee80211_csa_ie)
|
2011-11-08 04:00:24 +00:00
|
|
|
+ sizeof(struct ieee80211_quiet_ie)
|
2008-04-20 20:35:46 +00:00
|
|
|
+ 3
|
|
|
|
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE)
|
2008-07-07 17:24:55 +00:00
|
|
|
+ sizeof(struct ieee80211_ie_wpa)
|
2008-04-20 20:35:46 +00:00
|
|
|
+ sizeof(struct ieee80211_ie_htcap)
|
|
|
|
+ sizeof(struct ieee80211_ie_htinfo)
|
|
|
|
+ sizeof(struct ieee80211_ie_wpa)
|
|
|
|
+ sizeof(struct ieee80211_wme_param)
|
|
|
|
+ 4 + sizeof(struct ieee80211_ie_htcap)
|
|
|
|
+ 4 + sizeof(struct ieee80211_ie_htinfo)
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2008-04-20 20:35:46 +00:00
|
|
|
+ sizeof(struct ieee80211_ath_ie)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#endif
|
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
+ 2 + IEEE80211_MESHID_LEN
|
|
|
|
+ sizeof(struct ieee80211_meshconf_ie)
|
2009-03-24 20:39:08 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
+ (vap->iv_appie_proberesp != NULL ?
|
|
|
|
vap->iv_appie_proberesp->ie_len : 0)
|
|
|
|
);
|
|
|
|
if (m == NULL) {
|
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
memset(frm, 0, 8); /* timestamp should be filled later */
|
|
|
|
frm += 8;
|
|
|
|
*(uint16_t *)frm = htole16(bss->ni_intval);
|
|
|
|
frm += 2;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
capinfo = ieee80211_getcapinfo(vap, bss->ni_chan);
|
2008-04-20 20:35:46 +00:00
|
|
|
*(uint16_t *)frm = htole16(capinfo);
|
|
|
|
frm += 2;
|
|
|
|
|
|
|
|
frm = ieee80211_add_ssid(frm, bss->ni_essid, bss->ni_esslen);
|
|
|
|
rs = ieee80211_get_suprates(ic, bss->ni_chan);
|
|
|
|
frm = ieee80211_add_rates(frm, rs);
|
|
|
|
|
|
|
|
if (IEEE80211_IS_CHAN_FHSS(bss->ni_chan)) {
|
|
|
|
*frm++ = IEEE80211_ELEMID_FHPARMS;
|
|
|
|
*frm++ = 5;
|
|
|
|
*frm++ = bss->ni_fhdwell & 0x00ff;
|
|
|
|
*frm++ = (bss->ni_fhdwell >> 8) & 0x00ff;
|
|
|
|
*frm++ = IEEE80211_FH_CHANSET(
|
|
|
|
ieee80211_chan2ieee(ic, bss->ni_chan));
|
|
|
|
*frm++ = IEEE80211_FH_CHANPAT(
|
|
|
|
ieee80211_chan2ieee(ic, bss->ni_chan));
|
|
|
|
*frm++ = bss->ni_fhindex;
|
|
|
|
} else {
|
|
|
|
*frm++ = IEEE80211_ELEMID_DSPARMS;
|
|
|
|
*frm++ = 1;
|
|
|
|
*frm++ = ieee80211_chan2ieee(ic, bss->ni_chan);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_IBSS) {
|
|
|
|
*frm++ = IEEE80211_ELEMID_IBSSPARMS;
|
|
|
|
*frm++ = 2;
|
|
|
|
*frm++ = 0; *frm++ = 0; /* TODO: ATIM window */
|
|
|
|
}
|
|
|
|
if ((vap->iv_flags & IEEE80211_F_DOTH) ||
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_DOTD))
|
|
|
|
frm = ieee80211_add_countryie(frm, ic);
|
|
|
|
if (vap->iv_flags & IEEE80211_F_DOTH) {
|
|
|
|
if (IEEE80211_IS_CHAN_5GHZ(bss->ni_chan))
|
|
|
|
frm = ieee80211_add_powerconstraint(frm, vap);
|
|
|
|
if (ic->ic_flags & IEEE80211_F_CSAPENDING)
|
|
|
|
frm = ieee80211_add_csa(frm, vap);
|
|
|
|
}
|
2011-11-08 04:00:24 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_DOTH) {
|
|
|
|
if (IEEE80211_IS_CHAN_DFS(ic->ic_bsschan) &&
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_DFS)) {
|
|
|
|
if (vap->iv_quiet)
|
|
|
|
frm = ieee80211_add_quiet(frm, vap);
|
|
|
|
}
|
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
if (IEEE80211_IS_CHAN_ANYG(bss->ni_chan))
|
|
|
|
frm = ieee80211_add_erp(frm, ic);
|
|
|
|
frm = ieee80211_add_xrates(frm, rs);
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_rsn(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
/*
|
|
|
|
* NB: legacy 11b clients do not get certain ie's.
|
|
|
|
* The caller identifies such clients by passing
|
|
|
|
* a token in legacy to us. Could expand this to be
|
|
|
|
* any legacy client for stuff like HT ie's.
|
|
|
|
*/
|
|
|
|
if (IEEE80211_IS_CHAN_HT(bss->ni_chan) &&
|
|
|
|
legacy != IEEE80211_SEND_LEGACY_11B) {
|
|
|
|
frm = ieee80211_add_htcap(frm, bss);
|
|
|
|
frm = ieee80211_add_htinfo(frm, bss);
|
|
|
|
}
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_wpa(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_WME)
|
|
|
|
frm = ieee80211_add_wme_param(frm, &ic->ic_wme);
|
|
|
|
if (IEEE80211_IS_CHAN_HT(bss->ni_chan) &&
|
2009-06-07 22:00:22 +00:00
|
|
|
(vap->iv_flags_ht & IEEE80211_FHT_HTCOMPAT) &&
|
2008-04-20 20:35:46 +00:00
|
|
|
legacy != IEEE80211_SEND_LEGACY_11B) {
|
|
|
|
frm = ieee80211_add_htcap_vendor(frm, bss);
|
|
|
|
frm = ieee80211_add_htinfo_vendor(frm, bss);
|
|
|
|
}
|
2009-03-24 20:39:08 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2009-03-26 19:13:11 +00:00
|
|
|
if ((vap->iv_flags & IEEE80211_F_ATHEROS) &&
|
|
|
|
legacy != IEEE80211_SEND_LEGACY_11B)
|
|
|
|
frm = ieee80211_add_athcaps(frm, bss);
|
2009-03-24 20:39:08 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_appie_proberesp != NULL)
|
|
|
|
frm = add_appie(frm, vap->iv_appie_proberesp);
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS) {
|
|
|
|
frm = ieee80211_add_meshid(frm, vap);
|
|
|
|
frm = ieee80211_add_meshconf(frm, vap);
|
|
|
|
}
|
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = frm - mtod(m, uint8_t *);
|
|
|
|
|
|
|
|
return m;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Send a probe response frame to the specified mac address.
|
|
|
|
* This does not go through the normal mgt frame api so we
|
|
|
|
* can specify the destination address and re-use the bss node
|
|
|
|
* for the sta reference.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
ieee80211_send_proberesp(struct ieee80211vap *vap,
|
|
|
|
const uint8_t da[IEEE80211_ADDR_LEN], int legacy)
|
|
|
|
{
|
|
|
|
struct ieee80211_node *bss = vap->iv_bss;
|
|
|
|
struct ieee80211com *ic = vap->iv_ic;
|
|
|
|
struct ieee80211_frame *wh;
|
|
|
|
struct mbuf *m;
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
int ret;
|
2008-04-20 20:35:46 +00:00
|
|
|
|
|
|
|
if (vap->iv_state == IEEE80211_S_CAC) {
|
|
|
|
IEEE80211_NOTE(vap, IEEE80211_MSG_OUTPUT, bss,
|
|
|
|
"block %s frame in CAC state", "probe response");
|
|
|
|
vap->iv_stats.is_tx_badstate++;
|
|
|
|
return EIO; /* XXX */
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Hold a reference on the node so it doesn't go away until after
|
|
|
|
* the xmit is complete all the way in the driver. On error we
|
|
|
|
* will remove our reference.
|
|
|
|
*/
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_NODE,
|
|
|
|
"ieee80211_ref_node (%s:%u) %p<%s> refcnt %d\n",
|
|
|
|
__func__, __LINE__, bss, ether_sprintf(bss->ni_macaddr),
|
|
|
|
ieee80211_node_refcnt(bss)+1);
|
|
|
|
ieee80211_ref_node(bss);
|
|
|
|
|
|
|
|
m = ieee80211_alloc_proberesp(bss, legacy);
|
|
|
|
if (m == NULL) {
|
|
|
|
ieee80211_free_node(bss);
|
|
|
|
return ENOMEM;
|
|
|
|
}
|
|
|
|
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, sizeof(struct ieee80211_frame), M_NOWAIT);
|
2008-04-20 20:35:46 +00:00
|
|
|
KASSERT(m != NULL, ("no room for header"));
|
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
IEEE80211_TX_LOCK(ic);
|
2008-04-20 20:35:46 +00:00
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
2009-04-26 21:34:53 +00:00
|
|
|
ieee80211_send_setup(bss, m,
|
2008-10-26 00:43:11 +00:00
|
|
|
IEEE80211_FC0_TYPE_MGT | IEEE80211_FC0_SUBTYPE_PROBE_RESP,
|
|
|
|
IEEE80211_NONQOS_TID, vap->iv_myaddr, da, bss->ni_bssid);
|
2008-04-20 20:35:46 +00:00
|
|
|
/* XXX power management? */
|
2008-10-26 00:55:38 +00:00
|
|
|
m->m_flags |= M_ENCAP; /* mark encapsulated */
|
2008-04-20 20:35:46 +00:00
|
|
|
|
|
|
|
M_WME_SETAC(m, WME_AC_BE);
|
|
|
|
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_DEBUG | IEEE80211_MSG_DUMPPKTS,
|
|
|
|
"send probe resp on channel %u to %s%s\n",
|
|
|
|
ieee80211_chan2ieee(ic, ic->ic_curchan), ether_sprintf(da),
|
|
|
|
legacy ? " <legacy>" : "");
|
|
|
|
IEEE80211_NODE_STAT(bss, tx_mgmt);
|
|
|
|
|
Bring over my initial work from the net80211 TX locking branch.
This patchset implements a new TX lock, covering both the per-VAP (and
thus per-node) TX locking and the serialisation through to the underlying
physical device.
This implements the hard requirement that frames to the underlying physical
device are scheduled to the underlying device in the same order that they
are processed at the VAP layer. This includes adding extra encapsulation
state (such as sequence numbers and CCMP IV numbers.) Any order mismatch
here will result in dropped packets at the receiver.
There are multiple transmit contexts from the upper protocol layers as well
as the "raw" interface via the management and BPF transmit paths.
All of these need to be correctly serialised or bad behaviour will result
under load.
The specifics:
* add a new TX IC lock - it will eventually just be used for serialisation
to the underlying physical device but for now it's used for both the
VAP encapsulation/serialisation and the physical device dispatch.
This lock is specifically non-recursive.
* Methodize the parent transmit, vap transmit and ic_raw_xmit function
pointers; use lock assertions in the parent/vap transmit routines.
* Add a lock assertion in ieee80211_encap() - the TX lock must be held
here to guarantee sensible behaviour.
* Refactor out the packet sending code from ieee80211_start() - now
ieee80211_start() is just a loop over the ifnet queue and it dispatches
each VAP packet send through ieee80211_start_pkt().
Yes, I will likely rename ieee80211_start_pkt() to something that
better reflects its status as a VAP packet transmit path. More on
that later.
* Add locking around the management and BAR TX sending - to ensure that
encapsulation and TX are done hand-in-hand.
* Add locking in the mesh code - again, to ensure that encapsulation
and mesh transmit are done hand-in-hand.
* Add locking around the power save queue and ageq handling, when
dispatching to the parent interface.
* Add locking around the WDS handoff.
* Add a note in the mesh dispatch code that the TX path needs to be
re-thought-out - right now it's doing a direct parent device transmit
rather than going via the vap layer. It may "work", but it's likely
incorrect (as it bypasses any possible per-node power save and
aggregation handling.)
Why not a per-VAP or per-node lock?
Because in order to ensure per-VAP ordering, we'd have to hold the
VAP lock across parent->if_transmit(). There are a few problems
with this:
* There's some state being setup during each driver transmit - specifically,
the encryption encap / CCMP IV setup. That should eventually be dragged
back into the encapsulation phase but for now it lives in the driver TX path.
This should be locked.
* Two drivers (ath, iwn) re-use the node->ni_txseqs array in order to
allocate sequence numbers when doing transmit aggregation. This should
also be locked.
* Drivers may have multiple frames queued already - so when one calls
if_transmit(), it may end up dispatching multiple frames for different
VAPs/nodes, each needing a different lock when handling that particular
end destination.
So to be "correct" locking-wise, we'd end up needing to grab a VAP or
node lock inside the driver TX path when setting up crypto / AMPDU sequence
numbers, and we may already _have_ a TX lock held - mostly for the same
destination vap/node, but sometimes it'll be for others. That could lead
to LORs and thus deadlocks.
So for now, I'm sticking with an IC TX lock. It has the advantage of
papering over the above and it also has the added advantage that I can
assert that it's being held when doing a parent device transmit.
I'll look at splitting the locks out a bit more later on.
General outstanding net80211 TX path issues / TODO:
* Look into separating out the VAP serialisation and the IC handoff.
It's going to be tricky as parent->if_transmit() doesn't give me the
opportunity to split queuing from driver dispatch. See above.
* Work with monthadar to fix up the mesh transmit path so it doesn't go via
the parent interface when retransmitting frames.
* Push the encryption handling back into the driver, if it's at all
architectually sane to do so. I know it's possible - it's what mac80211
in Linux does.
* Make ieee80211_raw_xmit() queue a frame into VAP or parent queue rather
than doing a short-cut direct into the driver. There are QoS issues
here - you do want your management frames to be encapsulated and pushed
onto the stack sooner than the (large, bursty) amount of data frames
that are queued. But there has to be a saner way to do this.
* Fragments are still broken - drivers need to be upgraded to an if_transmit()
implementation and then fragmentation handling needs to be properly fixed.
Tested:
* STA - AR5416, AR9280, Intel 5300 abgn wifi
* Hostap - AR5416, AR9160, AR9280
* Mesh - some testing by monthadar@, more to come.
2013-03-08 20:23:55 +00:00
|
|
|
ret = ieee80211_raw_output(vap, bss, m, NULL);
|
|
|
|
IEEE80211_TX_UNLOCK(ic);
|
|
|
|
return (ret);
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate and build a RTS (Request To Send) control frame.
|
|
|
|
*/
|
|
|
|
struct mbuf *
|
|
|
|
ieee80211_alloc_rts(struct ieee80211com *ic,
|
|
|
|
const uint8_t ra[IEEE80211_ADDR_LEN],
|
|
|
|
const uint8_t ta[IEEE80211_ADDR_LEN],
|
|
|
|
uint16_t dur)
|
|
|
|
{
|
|
|
|
struct ieee80211_frame_rts *rts;
|
|
|
|
struct mbuf *m;
|
|
|
|
|
|
|
|
/* XXX honor ic_headroom */
|
2012-12-05 08:04:20 +00:00
|
|
|
m = m_gethdr(M_NOWAIT, MT_DATA);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (m != NULL) {
|
|
|
|
rts = mtod(m, struct ieee80211_frame_rts *);
|
|
|
|
rts->i_fc[0] = IEEE80211_FC0_VERSION_0 |
|
|
|
|
IEEE80211_FC0_TYPE_CTL | IEEE80211_FC0_SUBTYPE_RTS;
|
|
|
|
rts->i_fc[1] = IEEE80211_FC1_DIR_NODS;
|
|
|
|
*(u_int16_t *)rts->i_dur = htole16(dur);
|
|
|
|
IEEE80211_ADDR_COPY(rts->i_ra, ra);
|
|
|
|
IEEE80211_ADDR_COPY(rts->i_ta, ta);
|
|
|
|
|
|
|
|
m->m_pkthdr.len = m->m_len = sizeof(struct ieee80211_frame_rts);
|
|
|
|
}
|
|
|
|
return m;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate and build a CTS (Clear To Send) control frame.
|
|
|
|
*/
|
|
|
|
struct mbuf *
|
|
|
|
ieee80211_alloc_cts(struct ieee80211com *ic,
|
|
|
|
const uint8_t ra[IEEE80211_ADDR_LEN], uint16_t dur)
|
|
|
|
{
|
|
|
|
struct ieee80211_frame_cts *cts;
|
|
|
|
struct mbuf *m;
|
|
|
|
|
|
|
|
/* XXX honor ic_headroom */
|
2012-12-05 08:04:20 +00:00
|
|
|
m = m_gethdr(M_NOWAIT, MT_DATA);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (m != NULL) {
|
|
|
|
cts = mtod(m, struct ieee80211_frame_cts *);
|
|
|
|
cts->i_fc[0] = IEEE80211_FC0_VERSION_0 |
|
|
|
|
IEEE80211_FC0_TYPE_CTL | IEEE80211_FC0_SUBTYPE_CTS;
|
|
|
|
cts->i_fc[1] = IEEE80211_FC1_DIR_NODS;
|
|
|
|
*(u_int16_t *)cts->i_dur = htole16(dur);
|
|
|
|
IEEE80211_ADDR_COPY(cts->i_ra, ra);
|
|
|
|
|
|
|
|
m->m_pkthdr.len = m->m_len = sizeof(struct ieee80211_frame_cts);
|
|
|
|
}
|
|
|
|
return m;
|
|
|
|
}
|
|
|
|
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static void
|
|
|
|
ieee80211_tx_mgt_timeout(void *arg)
|
|
|
|
{
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
struct ieee80211vap *vap = arg;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
IEEE80211_LOCK(vap->iv_ic);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_state != IEEE80211_S_INIT &&
|
|
|
|
(vap->iv_ic->ic_flags & IEEE80211_F_SCAN) == 0) {
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
/*
|
|
|
|
* NB: it's safe to specify a timeout as the reason here;
|
|
|
|
* it'll only be used in the right state.
|
|
|
|
*/
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
ieee80211_new_state_locked(vap, IEEE80211_S_SCAN,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
IEEE80211_SCAN_FAIL_TIMEOUT);
|
|
|
|
}
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
IEEE80211_UNLOCK(vap->iv_ic);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
}
|
|
|
|
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
/*
|
|
|
|
* This is the callback set on net80211-sourced transmitted
|
|
|
|
* authentication request frames.
|
|
|
|
*
|
|
|
|
* This does a couple of things:
|
|
|
|
*
|
|
|
|
* + If the frame transmitted was a success, it schedules a future
|
|
|
|
* event which will transition the interface to scan.
|
|
|
|
* If a state transition _then_ occurs before that event occurs,
|
|
|
|
* said state transition will cancel this callout.
|
|
|
|
*
|
|
|
|
* + If the frame transmit was a failure, it immediately schedules
|
|
|
|
* the transition back to scan.
|
|
|
|
*/
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
static void
|
|
|
|
ieee80211_tx_mgt_cb(struct ieee80211_node *ni, void *arg, int status)
|
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
enum ieee80211_state ostate = (enum ieee80211_state) arg;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Frame transmit completed; arrange timer callback. If
|
|
|
|
* transmit was successfuly we wait for response. Otherwise
|
|
|
|
* we arrange an immediate callback instead of doing the
|
|
|
|
* callback directly since we don't know what state the driver
|
|
|
|
* is in (e.g. what locks it is holding). This work should
|
|
|
|
* not be too time-critical and not happen too often so the
|
|
|
|
* added overhead is acceptable.
|
|
|
|
*
|
|
|
|
* XXX what happens if !acked but response shows up before callback?
|
|
|
|
*/
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
if (vap->iv_state == ostate) {
|
2008-04-20 20:35:46 +00:00
|
|
|
callout_reset(&vap->iv_mgtsend,
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
status == 0 ? IEEE80211_TRANS_WAIT*hz : 0,
|
Fix a use-after-free node reference issue when waiting for a return
from a management frame transmission.
This bug is a bit loopy, so here goes.
The underlying cause is pretty easy to understand - the node isn't
referenced before passing into the callout, so if the node is deleted
before the callout fires, it'll dereference free'd memory.
The code path however is slightly more convoluted.
The functions _say_ mgt_tx - ie management transmit - which is partially
true. Yes, that callback is attached to the mbuf for some management
frames. However, it's only for frames relating to scanning and
authentication attempts. It helpfully drives the VAP state back to
"SCAN" if the transmission fails _OR_ (as I subsequently found out!)
if the transmission succeeds but the state machine doesn't make progress
towards being authenticated and active.
Now, the code itself isn't terribly clear about this.
It _looks_ like it's just handling the transmit failure case.
However, when you look at what goes on in the transmit success case, it's
moving the VAP state back to SCAN if it hasn't changed state since
the time the callback was scheduled. Ie, if it's in ASSOC or AUTH still,
it'll go back to SCAN. But if it has transitioned to the RUN state,
the comparison will fail and it'll not transition things back to the
SCAN state.
So, to fix this, I decided to leave everything the way it is and merely
fix the locking and remove the node reference.
The _better_ fix would be to turn this callout into a "assoc/auth request"
timeout callback and make the callout locked, thus eliminating all races.
However, until all the drivers have been fixed so that transmit completions
occur outside of any locking that's going on, it's going to be impossible
to do this without introducing LORs. So, I leave some of the evilness
in there.
Tested:
* AR5212, ath(4), STA mode
* 5100 and 4965 wifi, iwn(4), STA mode
2013-10-24 17:04:16 +00:00
|
|
|
ieee80211_tx_mgt_timeout, vap);
|
|
|
|
}
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
static void
|
|
|
|
ieee80211_beacon_construct(struct mbuf *m, uint8_t *frm,
|
|
|
|
struct ieee80211_beacon_offsets *bo, struct ieee80211_node *ni)
|
2004-12-08 17:26:47 +00:00
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
2007-09-17 19:07:24 +00:00
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211_rateset *rs = &ni->ni_rates;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
uint16_t capinfo;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* beacon frame format
|
|
|
|
* [8] time stamp
|
|
|
|
* [2] beacon interval
|
|
|
|
* [2] cabability information
|
|
|
|
* [tlv] ssid
|
|
|
|
* [tlv] supported rates
|
|
|
|
* [3] parameter set (DS)
|
2008-04-20 20:35:46 +00:00
|
|
|
* [8] CF parameter set (optional)
|
2004-12-08 17:26:47 +00:00
|
|
|
* [tlv] parameter set (IBSS/TIM)
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] country (optional)
|
|
|
|
* [3] power control (optional)
|
|
|
|
* [5] channel switch announcement (CSA) (optional)
|
2004-12-08 17:26:47 +00:00
|
|
|
* [tlv] extended rate phy (ERP)
|
|
|
|
* [tlv] extended supported rates
|
2008-07-07 17:24:55 +00:00
|
|
|
* [tlv] RSN parameters
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
* [tlv] HT capabilities
|
|
|
|
* [tlv] HT information
|
2008-04-20 20:35:46 +00:00
|
|
|
* XXX Vendor-specific OIDs (e.g. Atheros)
|
|
|
|
* [tlv] WPA parameters
|
|
|
|
* [tlv] WME parameters
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
* [tlv] Vendor OUI HT capabilities (optional)
|
|
|
|
* [tlv] Vendor OUI HT information (optional)
|
2009-03-26 19:13:11 +00:00
|
|
|
* [tlv] Atheros capabilities (optional)
|
2009-01-08 17:12:47 +00:00
|
|
|
* [tlv] TDMA parameters (optional)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
* [tlv] Mesh ID (MBSS)
|
|
|
|
* [tlv] Mesh Conf (MBSS)
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] application data (optional)
|
2004-12-08 17:26:47 +00:00
|
|
|
*/
|
|
|
|
|
2007-11-02 05:22:25 +00:00
|
|
|
memset(bo, 0, sizeof(*bo));
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
memset(frm, 0, 8); /* XXX timestamp is set by hardware/driver */
|
|
|
|
frm += 8;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)frm = htole16(ni->ni_intval);
|
2004-12-08 17:26:47 +00:00
|
|
|
frm += 2;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
capinfo = ieee80211_getcapinfo(vap, ni->ni_chan);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
bo->bo_caps = (uint16_t *)frm;
|
|
|
|
*(uint16_t *)frm = htole16(capinfo);
|
2004-12-08 17:26:47 +00:00
|
|
|
frm += 2;
|
|
|
|
*frm++ = IEEE80211_ELEMID_SSID;
|
2008-04-20 20:35:46 +00:00
|
|
|
if ((vap->iv_flags & IEEE80211_F_HIDESSID) == 0) {
|
2004-12-08 17:26:47 +00:00
|
|
|
*frm++ = ni->ni_esslen;
|
|
|
|
memcpy(frm, ni->ni_essid, ni->ni_esslen);
|
|
|
|
frm += ni->ni_esslen;
|
|
|
|
} else
|
|
|
|
*frm++ = 0;
|
|
|
|
frm = ieee80211_add_rates(frm, rs);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (!IEEE80211_IS_CHAN_FHSS(ni->ni_chan)) {
|
2004-12-08 17:26:47 +00:00
|
|
|
*frm++ = IEEE80211_ELEMID_DSPARMS;
|
|
|
|
*frm++ = 1;
|
2008-04-20 20:35:46 +00:00
|
|
|
*frm++ = ieee80211_chan2ieee(ic, ni->ni_chan);
|
|
|
|
}
|
|
|
|
if (ic->ic_flags & IEEE80211_F_PCF) {
|
|
|
|
bo->bo_cfp = frm;
|
|
|
|
frm = ieee80211_add_cfparms(frm, ic);
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
bo->bo_tim = frm;
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_opmode == IEEE80211_M_IBSS) {
|
2004-12-08 17:26:47 +00:00
|
|
|
*frm++ = IEEE80211_ELEMID_IBSSPARMS;
|
|
|
|
*frm++ = 2;
|
|
|
|
*frm++ = 0; *frm++ = 0; /* TODO: ATIM window */
|
|
|
|
bo->bo_tim_len = 0;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
} else if (vap->iv_opmode == IEEE80211_M_HOSTAP ||
|
|
|
|
vap->iv_opmode == IEEE80211_M_MBSS) {
|
|
|
|
/* TIM IE is the same for Mesh and Hostap */
|
2004-12-08 17:26:47 +00:00
|
|
|
struct ieee80211_tim_ie *tie = (struct ieee80211_tim_ie *) frm;
|
|
|
|
|
|
|
|
tie->tim_ie = IEEE80211_ELEMID_TIM;
|
|
|
|
tie->tim_len = 4; /* length */
|
|
|
|
tie->tim_count = 0; /* DTIM count */
|
2008-04-20 20:35:46 +00:00
|
|
|
tie->tim_period = vap->iv_dtim_period; /* DTIM period */
|
2004-12-08 17:26:47 +00:00
|
|
|
tie->tim_bitctl = 0; /* bitmap control */
|
|
|
|
tie->tim_bitmap[0] = 0; /* Partial Virtual Bitmap */
|
|
|
|
frm += sizeof(struct ieee80211_tim_ie);
|
|
|
|
bo->bo_tim_len = 1;
|
|
|
|
}
|
2007-09-17 19:07:24 +00:00
|
|
|
bo->bo_tim_trailer = frm;
|
2008-04-20 20:35:46 +00:00
|
|
|
if ((vap->iv_flags & IEEE80211_F_DOTH) ||
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_DOTD))
|
|
|
|
frm = ieee80211_add_countryie(frm, ic);
|
|
|
|
if (vap->iv_flags & IEEE80211_F_DOTH) {
|
|
|
|
if (IEEE80211_IS_CHAN_5GHZ(ni->ni_chan))
|
|
|
|
frm = ieee80211_add_powerconstraint(frm, vap);
|
|
|
|
bo->bo_csa = frm;
|
|
|
|
if (ic->ic_flags & IEEE80211_F_CSAPENDING)
|
2011-11-08 04:00:24 +00:00
|
|
|
frm = ieee80211_add_csa(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
} else
|
|
|
|
bo->bo_csa = frm;
|
2011-11-08 04:00:24 +00:00
|
|
|
|
|
|
|
if (vap->iv_flags & IEEE80211_F_DOTH) {
|
|
|
|
bo->bo_quiet = frm;
|
|
|
|
if (IEEE80211_IS_CHAN_DFS(ic->ic_bsschan) &&
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_DFS)) {
|
|
|
|
if (vap->iv_quiet)
|
|
|
|
frm = ieee80211_add_quiet(frm,vap);
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
bo->bo_quiet = frm;
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (IEEE80211_IS_CHAN_ANYG(ni->ni_chan)) {
|
2006-01-02 16:57:20 +00:00
|
|
|
bo->bo_erp = frm;
|
2004-12-08 17:26:47 +00:00
|
|
|
frm = ieee80211_add_erp(frm, ic);
|
2007-11-02 05:22:25 +00:00
|
|
|
}
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
frm = ieee80211_add_xrates(frm, rs);
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_rsn(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (IEEE80211_IS_CHAN_HT(ni->ni_chan)) {
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
frm = ieee80211_add_htcap(frm, ni);
|
|
|
|
bo->bo_htinfo = frm;
|
|
|
|
frm = ieee80211_add_htinfo(frm, ni);
|
2007-11-02 05:22:25 +00:00
|
|
|
}
|
2013-07-25 06:23:26 +00:00
|
|
|
frm = ieee80211_add_wpa(frm, vap);
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_WME) {
|
2007-11-02 05:22:25 +00:00
|
|
|
bo->bo_wme = frm;
|
|
|
|
frm = ieee80211_add_wme_param(frm, &ic->ic_wme);
|
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
if (IEEE80211_IS_CHAN_HT(ni->ni_chan) &&
|
2009-06-07 22:00:22 +00:00
|
|
|
(vap->iv_flags_ht & IEEE80211_FHT_HTCOMPAT)) {
|
2007-11-02 05:22:25 +00:00
|
|
|
frm = ieee80211_add_htcap_vendor(frm, ni);
|
|
|
|
frm = ieee80211_add_htinfo_vendor(frm, ni);
|
|
|
|
}
|
2009-03-26 19:13:11 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
|
|
|
if (vap->iv_flags & IEEE80211_F_ATHEROS) {
|
|
|
|
bo->bo_ath = frm;
|
|
|
|
frm = ieee80211_add_athcaps(frm, ni);
|
|
|
|
}
|
|
|
|
#endif
|
2009-01-08 17:12:47 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_TDMA
|
|
|
|
if (vap->iv_caps & IEEE80211_C_TDMA) {
|
|
|
|
bo->bo_tdma = frm;
|
|
|
|
frm = ieee80211_add_tdma(frm, vap);
|
|
|
|
}
|
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_appie_beacon != NULL) {
|
|
|
|
bo->bo_appie = frm;
|
|
|
|
bo->bo_appie_len = vap->iv_appie_beacon->ie_len;
|
|
|
|
frm = add_appie(frm, vap->iv_appie_beacon);
|
|
|
|
}
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS) {
|
|
|
|
frm = ieee80211_add_meshid(frm, vap);
|
2009-10-19 18:46:22 +00:00
|
|
|
bo->bo_meshconf = frm;
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
frm = ieee80211_add_meshconf(frm, vap);
|
|
|
|
}
|
|
|
|
#endif
|
2007-09-17 19:07:24 +00:00
|
|
|
bo->bo_tim_trailer_len = frm - bo->bo_tim_trailer;
|
2008-04-20 20:35:46 +00:00
|
|
|
bo->bo_csa_trailer_len = frm - bo->bo_csa;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
m->m_pkthdr.len = m->m_len = frm - mtod(m, uint8_t *);
|
2008-04-20 20:35:46 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate a beacon frame and fillin the appropriate bits.
|
|
|
|
*/
|
|
|
|
struct mbuf *
|
|
|
|
ieee80211_beacon_alloc(struct ieee80211_node *ni,
|
|
|
|
struct ieee80211_beacon_offsets *bo)
|
|
|
|
{
|
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
|
|
|
struct ifnet *ifp = vap->iv_ifp;
|
|
|
|
struct ieee80211_frame *wh;
|
|
|
|
struct mbuf *m;
|
|
|
|
int pktlen;
|
|
|
|
uint8_t *frm;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* beacon frame format
|
|
|
|
* [8] time stamp
|
|
|
|
* [2] beacon interval
|
|
|
|
* [2] cabability information
|
|
|
|
* [tlv] ssid
|
|
|
|
* [tlv] supported rates
|
|
|
|
* [3] parameter set (DS)
|
|
|
|
* [8] CF parameter set (optional)
|
|
|
|
* [tlv] parameter set (IBSS/TIM)
|
|
|
|
* [tlv] country (optional)
|
|
|
|
* [3] power control (optional)
|
|
|
|
* [5] channel switch announcement (CSA) (optional)
|
|
|
|
* [tlv] extended rate phy (ERP)
|
|
|
|
* [tlv] extended supported rates
|
|
|
|
* [tlv] RSN parameters
|
|
|
|
* [tlv] HT capabilities
|
|
|
|
* [tlv] HT information
|
|
|
|
* [tlv] Vendor OUI HT capabilities (optional)
|
|
|
|
* [tlv] Vendor OUI HT information (optional)
|
|
|
|
* XXX Vendor-specific OIDs (e.g. Atheros)
|
|
|
|
* [tlv] WPA parameters
|
|
|
|
* [tlv] WME parameters
|
2009-01-08 17:12:47 +00:00
|
|
|
* [tlv] TDMA parameters (optional)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
* [tlv] Mesh ID (MBSS)
|
|
|
|
* [tlv] Mesh Conf (MBSS)
|
2008-04-20 20:35:46 +00:00
|
|
|
* [tlv] application data (optional)
|
|
|
|
* NB: we allocate the max space required for the TIM bitmap.
|
|
|
|
* XXX how big is this?
|
|
|
|
*/
|
|
|
|
pktlen = 8 /* time stamp */
|
|
|
|
+ sizeof(uint16_t) /* beacon interval */
|
|
|
|
+ sizeof(uint16_t) /* capabilities */
|
|
|
|
+ 2 + ni->ni_esslen /* ssid */
|
|
|
|
+ 2 + IEEE80211_RATE_SIZE /* supported rates */
|
|
|
|
+ 2 + 1 /* DS parameters */
|
|
|
|
+ 2 + 6 /* CF parameters */
|
|
|
|
+ 2 + 4 + vap->iv_tim_len /* DTIM/IBSSPARMS */
|
|
|
|
+ IEEE80211_COUNTRY_MAX_SIZE /* country */
|
|
|
|
+ 2 + 1 /* power control */
|
2011-11-08 04:00:24 +00:00
|
|
|
+ sizeof(struct ieee80211_csa_ie) /* CSA */
|
|
|
|
+ sizeof(struct ieee80211_quiet_ie) /* Quiet */
|
2008-04-20 20:35:46 +00:00
|
|
|
+ 2 + 1 /* ERP */
|
|
|
|
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE)
|
|
|
|
+ (vap->iv_caps & IEEE80211_C_WPA ? /* WPA 1+2 */
|
|
|
|
2*sizeof(struct ieee80211_ie_wpa) : 0)
|
|
|
|
/* XXX conditional? */
|
|
|
|
+ 4+2*sizeof(struct ieee80211_ie_htcap)/* HT caps */
|
|
|
|
+ 4+2*sizeof(struct ieee80211_ie_htinfo)/* HT info */
|
|
|
|
+ (vap->iv_caps & IEEE80211_C_WME ? /* WME */
|
|
|
|
sizeof(struct ieee80211_wme_param) : 0)
|
2009-03-26 19:13:11 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
|
|
|
+ sizeof(struct ieee80211_ath_ie) /* ATH */
|
|
|
|
#endif
|
2009-01-08 17:12:47 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_TDMA
|
|
|
|
+ (vap->iv_caps & IEEE80211_C_TDMA ? /* TDMA */
|
|
|
|
sizeof(struct ieee80211_tdma_param) : 0)
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
#endif
|
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
+ 2 + ni->ni_meshidlen
|
|
|
|
+ sizeof(struct ieee80211_meshconf_ie)
|
2009-01-08 17:12:47 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
+ IEEE80211_MAX_APPIE
|
|
|
|
;
|
|
|
|
m = ieee80211_getmgtframe(&frm,
|
|
|
|
ic->ic_headroom + sizeof(struct ieee80211_frame), pktlen);
|
|
|
|
if (m == NULL) {
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_ANY,
|
|
|
|
"%s: cannot get buf; size %u\n", __func__, pktlen);
|
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
ieee80211_beacon_construct(m, frm, bo, ni);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2012-12-05 08:04:20 +00:00
|
|
|
M_PREPEND(m, sizeof(struct ieee80211_frame), M_NOWAIT);
|
2004-12-08 17:26:47 +00:00
|
|
|
KASSERT(m != NULL, ("no space for 802.11 header?"));
|
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
|
|
|
wh->i_fc[0] = IEEE80211_FC0_VERSION_0 | IEEE80211_FC0_TYPE_MGT |
|
|
|
|
IEEE80211_FC0_SUBTYPE_BEACON;
|
|
|
|
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)wh->i_dur = 0;
|
2004-12-08 17:26:47 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr1, ifp->if_broadcastaddr);
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr2, vap->iv_myaddr);
|
2009-10-23 11:13:08 +00:00
|
|
|
IEEE80211_ADDR_COPY(wh->i_addr3, ni->ni_bssid);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
*(uint16_t *)wh->i_seq = 0;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
return m;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Update the dynamic parts of a beacon frame based on the current state.
|
|
|
|
*/
|
|
|
|
int
|
2007-09-17 19:07:24 +00:00
|
|
|
ieee80211_beacon_update(struct ieee80211_node *ni,
|
2004-12-08 17:26:47 +00:00
|
|
|
struct ieee80211_beacon_offsets *bo, struct mbuf *m, int mcast)
|
|
|
|
{
|
2008-04-20 20:35:46 +00:00
|
|
|
struct ieee80211vap *vap = ni->ni_vap;
|
2007-09-17 19:07:24 +00:00
|
|
|
struct ieee80211com *ic = ni->ni_ic;
|
2004-12-08 17:26:47 +00:00
|
|
|
int len_changed = 0;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
uint16_t capinfo;
|
2011-08-24 08:53:33 +00:00
|
|
|
struct ieee80211_frame *wh;
|
|
|
|
ieee80211_seq seqno;
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_LOCK(ic);
|
|
|
|
/*
|
|
|
|
* Handle 11h channel change when we've reached the count.
|
|
|
|
* We must recalculate the beacon frame contents to account
|
|
|
|
* for the new channel. Note we do this only for the first
|
|
|
|
* vap that reaches this point; subsequent vaps just update
|
|
|
|
* their beacon state to reflect the recalculated channel.
|
|
|
|
*/
|
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_CSA) &&
|
|
|
|
vap->iv_csa_count == ic->ic_csa_count) {
|
|
|
|
vap->iv_csa_count = 0;
|
|
|
|
/*
|
|
|
|
* Effect channel change before reconstructing the beacon
|
|
|
|
* frame contents as many places reference ni_chan.
|
|
|
|
*/
|
|
|
|
if (ic->ic_csa_newchan != NULL)
|
|
|
|
ieee80211_csa_completeswitch(ic);
|
|
|
|
/*
|
|
|
|
* NB: ieee80211_beacon_construct clears all pending
|
|
|
|
* updates in bo_flags so we don't need to explicitly
|
|
|
|
* clear IEEE80211_BEACON_CSA.
|
|
|
|
*/
|
|
|
|
ieee80211_beacon_construct(m,
|
|
|
|
mtod(m, uint8_t*) + sizeof(struct ieee80211_frame), bo, ni);
|
|
|
|
|
|
|
|
/* XXX do WME aggressive mode processing? */
|
|
|
|
IEEE80211_UNLOCK(ic);
|
|
|
|
return 1; /* just assume length changed */
|
|
|
|
}
|
|
|
|
|
2011-08-24 08:53:33 +00:00
|
|
|
wh = mtod(m, struct ieee80211_frame *);
|
|
|
|
seqno = ni->ni_txseqs[IEEE80211_NONQOS_TID]++;
|
|
|
|
*(uint16_t *)&wh->i_seq[0] =
|
|
|
|
htole16(seqno << IEEE80211_SEQ_SEQ_SHIFT);
|
|
|
|
M_SEQNO_SET(m, seqno);
|
|
|
|
|
2004-12-08 17:26:47 +00:00
|
|
|
/* XXX faster to recalculate entirely or just changes? */
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
capinfo = ieee80211_getcapinfo(vap, ni->ni_chan);
|
2004-12-08 17:26:47 +00:00
|
|
|
*bo->bo_caps = htole16(capinfo);
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_flags & IEEE80211_F_WME) {
|
2004-12-08 17:26:47 +00:00
|
|
|
struct ieee80211_wme_state *wme = &ic->ic_wme;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check for agressive mode change. When there is
|
|
|
|
* significant high priority traffic in the BSS
|
|
|
|
* throttle back BE traffic by using conservative
|
|
|
|
* parameters. Otherwise BE uses agressive params
|
|
|
|
* to optimize performance of legacy/non-QoS traffic.
|
|
|
|
*/
|
|
|
|
if (wme->wme_flags & WME_F_AGGRMODE) {
|
|
|
|
if (wme->wme_hipri_traffic >
|
|
|
|
wme->wme_hipri_switch_thresh) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_WME,
|
2004-12-08 17:26:47 +00:00
|
|
|
"%s: traffic %u, disable aggressive mode\n",
|
|
|
|
__func__, wme->wme_hipri_traffic);
|
|
|
|
wme->wme_flags &= ~WME_F_AGGRMODE;
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_wme_updateparams_locked(vap);
|
2004-12-08 17:26:47 +00:00
|
|
|
wme->wme_hipri_traffic =
|
|
|
|
wme->wme_hipri_switch_hysteresis;
|
|
|
|
} else
|
|
|
|
wme->wme_hipri_traffic = 0;
|
|
|
|
} else {
|
|
|
|
if (wme->wme_hipri_traffic <=
|
|
|
|
wme->wme_hipri_switch_thresh) {
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_WME,
|
2004-12-08 17:26:47 +00:00
|
|
|
"%s: traffic %u, enable aggressive mode\n",
|
|
|
|
__func__, wme->wme_hipri_traffic);
|
|
|
|
wme->wme_flags |= WME_F_AGGRMODE;
|
2008-04-20 20:35:46 +00:00
|
|
|
ieee80211_wme_updateparams_locked(vap);
|
2004-12-08 17:26:47 +00:00
|
|
|
wme->wme_hipri_traffic = 0;
|
|
|
|
} else
|
|
|
|
wme->wme_hipri_traffic =
|
|
|
|
wme->wme_hipri_switch_hysteresis;
|
|
|
|
}
|
2007-09-17 19:07:24 +00:00
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_WME)) {
|
2004-12-08 17:26:47 +00:00
|
|
|
(void) ieee80211_add_wme_param(bo->bo_wme, wme);
|
2007-09-17 19:07:24 +00:00
|
|
|
clrbit(bo->bo_flags, IEEE80211_BEACON_WME);
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_HTINFO)) {
|
|
|
|
ieee80211_ht_update_beacon(vap, bo);
|
2007-09-17 19:07:24 +00:00
|
|
|
clrbit(bo->bo_flags, IEEE80211_BEACON_HTINFO);
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
}
|
2009-01-08 17:12:47 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_TDMA
|
|
|
|
if (vap->iv_caps & IEEE80211_C_TDMA) {
|
|
|
|
/*
|
|
|
|
* NB: the beacon is potentially updated every TBTT.
|
|
|
|
*/
|
|
|
|
ieee80211_tdma_update_beacon(vap, bo);
|
|
|
|
}
|
|
|
|
#endif
|
2009-10-19 18:46:22 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
|
|
|
if (vap->iv_opmode == IEEE80211_M_MBSS)
|
|
|
|
ieee80211_mesh_update_beacon(vap, bo);
|
|
|
|
#endif
|
|
|
|
|
Implementation of the upcoming Wireless Mesh standard, 802.11s, on the
net80211 wireless stack. This work is based on the March 2009 D3.0 draft
standard. This standard is expected to become final next year.
This includes two main net80211 modules, ieee80211_mesh.c
which deals with peer link management, link metric calculation,
routing table control and mesh configuration and ieee80211_hwmp.c
which deals with the actually routing process on the mesh network.
HWMP is the mandatory routing protocol on by the mesh standard, but
others, such as RA-OLSR, can be implemented.
Authentication and encryption are not implemented.
There are several scripts under tools/tools/net80211/scripts that can be
used to test different mesh network topologies and they also teach you
how to setup a mesh vap (for the impatient: ifconfig wlan0 create
wlandev ... wlanmode mesh).
A new build option is available: IEEE80211_SUPPORT_MESH and it's enabled
by default on GENERIC kernels for i386, amd64, sparc64 and pc98.
Drivers that support mesh networks right now are: ath, ral and mwl.
More information at: http://wiki.freebsd.org/WifiMesh
Please note that this work is experimental. Also, please note that
bridging a mesh vap with another network interface is not yet supported.
Many thanks to the FreeBSD Foundation for sponsoring this project and to
Sam Leffler for his support.
Also, I would like to thank Gateworks Corporation for sending me a
Cambria board which was used during the development of this project.
Reviewed by: sam
Approved by: re (kensmith)
Obtained from: projects/mesh11s
2009-07-11 15:02:45 +00:00
|
|
|
if (vap->iv_opmode == IEEE80211_M_HOSTAP ||
|
|
|
|
vap->iv_opmode == IEEE80211_M_MBSS) { /* NB: no IBSS support*/
|
2004-12-08 17:26:47 +00:00
|
|
|
struct ieee80211_tim_ie *tie =
|
|
|
|
(struct ieee80211_tim_ie *) bo->bo_tim;
|
2007-09-17 19:07:24 +00:00
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_TIM)) {
|
2004-12-08 17:26:47 +00:00
|
|
|
u_int timlen, timoff, i;
|
|
|
|
/*
|
|
|
|
* ATIM/DTIM needs updating. If it fits in the
|
|
|
|
* current space allocated then just copy in the
|
|
|
|
* new bits. Otherwise we need to move any trailing
|
|
|
|
* data to make room. Note that we know there is
|
|
|
|
* contiguous space because ieee80211_beacon_allocate
|
|
|
|
* insures there is space in the mbuf to write a
|
2008-04-20 20:35:46 +00:00
|
|
|
* maximal-size virtual bitmap (based on iv_max_aid).
|
2004-12-08 17:26:47 +00:00
|
|
|
*/
|
|
|
|
/*
|
|
|
|
* Calculate the bitmap size and offset, copy any
|
|
|
|
* trailer out of the way, and then copy in the
|
|
|
|
* new bitmap and update the information element.
|
|
|
|
* Note that the tim bitmap must contain at least
|
|
|
|
* one byte and any offset must be even.
|
|
|
|
*/
|
2008-04-20 20:35:46 +00:00
|
|
|
if (vap->iv_ps_pending != 0) {
|
2004-12-08 17:26:47 +00:00
|
|
|
timoff = 128; /* impossibly large */
|
2008-04-20 20:35:46 +00:00
|
|
|
for (i = 0; i < vap->iv_tim_len; i++)
|
|
|
|
if (vap->iv_tim_bitmap[i]) {
|
2004-12-08 17:26:47 +00:00
|
|
|
timoff = i &~ 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
KASSERT(timoff != 128, ("tim bitmap empty!"));
|
2008-04-20 20:35:46 +00:00
|
|
|
for (i = vap->iv_tim_len-1; i >= timoff; i--)
|
|
|
|
if (vap->iv_tim_bitmap[i])
|
2004-12-08 17:26:47 +00:00
|
|
|
break;
|
|
|
|
timlen = 1 + (i - timoff);
|
|
|
|
} else {
|
|
|
|
timoff = 0;
|
|
|
|
timlen = 1;
|
|
|
|
}
|
|
|
|
if (timlen != bo->bo_tim_len) {
|
|
|
|
/* copy up/down trailer */
|
2006-01-02 16:57:20 +00:00
|
|
|
int adjust = tie->tim_bitmap+timlen
|
2007-09-17 19:07:24 +00:00
|
|
|
- bo->bo_tim_trailer;
|
|
|
|
ovbcopy(bo->bo_tim_trailer,
|
|
|
|
bo->bo_tim_trailer+adjust,
|
|
|
|
bo->bo_tim_trailer_len);
|
|
|
|
bo->bo_tim_trailer += adjust;
|
2006-01-02 16:57:20 +00:00
|
|
|
bo->bo_erp += adjust;
|
Update 802.11 wireless support:
o major overhaul of the way channels are handled: channels are now
fully enumerated and uniquely identify the operating characteristics;
these changes are visible to user applications which require changes
o make scanning support independent of the state machine to enable
background scanning and roaming
o move scanning support into loadable modules based on the operating
mode to enable different policies and reduce the memory footprint
on systems w/ constrained resources
o add background scanning in station mode (no support for adhoc/ibss
mode yet)
o significantly speedup sta mode scanning with a variety of techniques
o add roaming support when background scanning is supported; for now
we use a simple algorithm to trigger a roam: we threshold the rssi
and tx rate, if either drops too low we try to roam to a new ap
o add tx fragmentation support
o add first cut at 802.11n support: this code works with forthcoming
drivers but is incomplete; it's included now to establish a baseline
for other drivers to be developed and for user applications
o adjust max_linkhdr et. al. to reflect 802.11 requirements; this eliminates
prepending mbufs for traffic generated locally
o add support for Atheros protocol extensions; mainly the fast frames
encapsulation (note this can be used with any card that can tx+rx
large frames correctly)
o add sta support for ap's that beacon both WPA1+2 support
o change all data types from bsd-style to posix-style
o propagate noise floor data from drivers to net80211 and on to user apps
o correct various issues in the sta mode state machine related to handling
authentication and association failures
o enable the addition of sta mode power save support for drivers that need
net80211 support (not in this commit)
o remove old WI compatibility ioctls (wicontrol is officially dead)
o change the data structures returned for get sta info and get scan
results so future additions will not break user apps
o fixed tx rate is now maintained internally as an ieee rate and not an
index into the rate set; this needs to be extended to deal with
multi-mode operation
o add extended channel specifications to radiotap to enable 11n sniffing
Drivers:
o ath: add support for bg scanning, tx fragmentation, fast frames,
dynamic turbo (lightly tested), 11n (sniffing only and needs
new hal)
o awi: compile tested only
o ndis: lightly tested
o ipw: lightly tested
o iwi: add support for bg scanning (well tested but may have some
rough edges)
o ral, ural, rum: add suppoort for bg scanning, calibrate rssi data
o wi: lightly tested
This work is based on contributions by Atheros, kmacy, sephe, thompsa,
mlaier, kevlo, and others. Much of the scanning work was supported by
Atheros. The 11n work was supported by Marvell.
2007-06-11 03:36:55 +00:00
|
|
|
bo->bo_htinfo += adjust;
|
2011-02-21 20:02:02 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2009-03-26 19:13:11 +00:00
|
|
|
bo->bo_ath += adjust;
|
|
|
|
#endif
|
2011-02-21 20:02:02 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_TDMA
|
2009-03-26 19:04:11 +00:00
|
|
|
bo->bo_tdma += adjust;
|
2009-10-19 18:46:22 +00:00
|
|
|
#endif
|
2011-02-21 20:02:02 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
2009-10-19 18:46:22 +00:00
|
|
|
bo->bo_meshconf += adjust;
|
2009-03-26 19:04:11 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
bo->bo_appie += adjust;
|
|
|
|
bo->bo_wme += adjust;
|
|
|
|
bo->bo_csa += adjust;
|
2011-11-08 04:00:24 +00:00
|
|
|
bo->bo_quiet += adjust;
|
2004-12-08 17:26:47 +00:00
|
|
|
bo->bo_tim_len = timlen;
|
|
|
|
|
|
|
|
/* update information element */
|
|
|
|
tie->tim_len = 3 + timlen;
|
|
|
|
tie->tim_bitctl = timoff;
|
|
|
|
len_changed = 1;
|
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
memcpy(tie->tim_bitmap, vap->iv_tim_bitmap + timoff,
|
2004-12-08 17:26:47 +00:00
|
|
|
bo->bo_tim_len);
|
|
|
|
|
2007-09-17 19:07:24 +00:00
|
|
|
clrbit(bo->bo_flags, IEEE80211_BEACON_TIM);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
2008-04-20 20:35:46 +00:00
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_POWER,
|
2004-12-08 17:26:47 +00:00
|
|
|
"%s: TIM updated, pending %u, off %u, len %u\n",
|
2008-04-20 20:35:46 +00:00
|
|
|
__func__, vap->iv_ps_pending, timoff, timlen);
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
|
|
|
/* count down DTIM period */
|
|
|
|
if (tie->tim_count == 0)
|
|
|
|
tie->tim_count = tie->tim_period - 1;
|
|
|
|
else
|
|
|
|
tie->tim_count--;
|
|
|
|
/* update state for buffered multicast frames on DTIM */
|
2005-12-05 22:14:13 +00:00
|
|
|
if (mcast && tie->tim_count == 0)
|
2004-12-08 17:26:47 +00:00
|
|
|
tie->tim_bitctl |= 1;
|
|
|
|
else
|
|
|
|
tie->tim_bitctl &= ~1;
|
2008-04-20 20:35:46 +00:00
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_CSA)) {
|
|
|
|
struct ieee80211_csa_ie *csa =
|
|
|
|
(struct ieee80211_csa_ie *) bo->bo_csa;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Insert or update CSA ie. If we're just starting
|
|
|
|
* to count down to the channel switch then we need
|
|
|
|
* to insert the CSA ie. Otherwise we just need to
|
|
|
|
* drop the count. The actual change happens above
|
|
|
|
* when the vap's count reaches the target count.
|
|
|
|
*/
|
|
|
|
if (vap->iv_csa_count == 0) {
|
|
|
|
memmove(&csa[1], csa, bo->bo_csa_trailer_len);
|
|
|
|
bo->bo_erp += sizeof(*csa);
|
2009-03-26 19:06:00 +00:00
|
|
|
bo->bo_htinfo += sizeof(*csa);
|
2008-04-20 20:35:46 +00:00
|
|
|
bo->bo_wme += sizeof(*csa);
|
2011-02-21 20:02:02 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
2009-03-26 19:13:11 +00:00
|
|
|
bo->bo_ath += sizeof(*csa);
|
|
|
|
#endif
|
2011-02-21 20:02:02 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_TDMA
|
2009-03-26 19:04:11 +00:00
|
|
|
bo->bo_tdma += sizeof(*csa);
|
2009-10-19 18:46:22 +00:00
|
|
|
#endif
|
2011-02-21 20:02:02 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_MESH
|
2009-10-19 18:46:22 +00:00
|
|
|
bo->bo_meshconf += sizeof(*csa);
|
2009-03-26 19:04:11 +00:00
|
|
|
#endif
|
2008-04-20 20:35:46 +00:00
|
|
|
bo->bo_appie += sizeof(*csa);
|
|
|
|
bo->bo_csa_trailer_len += sizeof(*csa);
|
2011-11-08 04:00:24 +00:00
|
|
|
bo->bo_quiet += sizeof(*csa);
|
2008-04-20 20:35:46 +00:00
|
|
|
bo->bo_tim_trailer_len += sizeof(*csa);
|
|
|
|
m->m_len += sizeof(*csa);
|
|
|
|
m->m_pkthdr.len += sizeof(*csa);
|
|
|
|
|
|
|
|
ieee80211_add_csa(bo->bo_csa, vap);
|
|
|
|
} else
|
|
|
|
csa->csa_count--;
|
|
|
|
vap->iv_csa_count++;
|
|
|
|
/* NB: don't clear IEEE80211_BEACON_CSA */
|
|
|
|
}
|
2011-11-08 04:00:24 +00:00
|
|
|
if (IEEE80211_IS_CHAN_DFS(ic->ic_bsschan) &&
|
|
|
|
(vap->iv_flags_ext & IEEE80211_FEXT_DFS) ){
|
|
|
|
if (vap->iv_quiet)
|
|
|
|
ieee80211_add_quiet(bo->bo_quiet, vap);
|
|
|
|
}
|
2007-09-17 19:07:24 +00:00
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_ERP)) {
|
2006-01-02 16:57:20 +00:00
|
|
|
/*
|
|
|
|
* ERP element needs updating.
|
|
|
|
*/
|
|
|
|
(void) ieee80211_add_erp(bo->bo_erp, ic);
|
2007-09-17 19:07:24 +00:00
|
|
|
clrbit(bo->bo_flags, IEEE80211_BEACON_ERP);
|
2006-01-02 16:57:20 +00:00
|
|
|
}
|
2009-03-26 19:13:11 +00:00
|
|
|
#ifdef IEEE80211_SUPPORT_SUPERG
|
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_ATH)) {
|
|
|
|
ieee80211_add_athcaps(bo->bo_ath, ni);
|
|
|
|
clrbit(bo->bo_flags, IEEE80211_BEACON_ATH);
|
|
|
|
}
|
|
|
|
#endif
|
2004-12-08 17:26:47 +00:00
|
|
|
}
|
2008-04-20 20:35:46 +00:00
|
|
|
if (isset(bo->bo_flags, IEEE80211_BEACON_APPIE)) {
|
|
|
|
const struct ieee80211_appie *aie = vap->iv_appie_beacon;
|
|
|
|
int aielen;
|
|
|
|
uint8_t *frm;
|
|
|
|
|
|
|
|
aielen = 0;
|
|
|
|
if (aie != NULL)
|
|
|
|
aielen += aie->ie_len;
|
|
|
|
if (aielen != bo->bo_appie_len) {
|
|
|
|
/* copy up/down trailer */
|
|
|
|
int adjust = aielen - bo->bo_appie_len;
|
|
|
|
ovbcopy(bo->bo_tim_trailer, bo->bo_tim_trailer+adjust,
|
|
|
|
bo->bo_tim_trailer_len);
|
|
|
|
bo->bo_tim_trailer += adjust;
|
|
|
|
bo->bo_appie += adjust;
|
|
|
|
bo->bo_appie_len = aielen;
|
|
|
|
|
|
|
|
len_changed = 1;
|
|
|
|
}
|
|
|
|
frm = bo->bo_appie;
|
|
|
|
if (aie != NULL)
|
|
|
|
frm = add_appie(frm, aie);
|
|
|
|
clrbit(bo->bo_flags, IEEE80211_BEACON_APPIE);
|
|
|
|
}
|
|
|
|
IEEE80211_UNLOCK(ic);
|
2004-12-08 17:26:47 +00:00
|
|
|
|
|
|
|
return len_changed;
|
|
|
|
}
|
2013-08-26 09:52:05 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Do Ethernet-LLC encapsulation for each payload in a fast frame
|
|
|
|
* tunnel encapsulation. The frame is assumed to have an Ethernet
|
|
|
|
* header at the front that must be stripped before prepending the
|
|
|
|
* LLC followed by the Ethernet header passed in (with an Ethernet
|
|
|
|
* type that specifies the payload size).
|
|
|
|
*/
|
|
|
|
struct mbuf *
|
|
|
|
ieee80211_ff_encap1(struct ieee80211vap *vap, struct mbuf *m,
|
|
|
|
const struct ether_header *eh)
|
|
|
|
{
|
|
|
|
struct llc *llc;
|
|
|
|
uint16_t payload;
|
|
|
|
|
|
|
|
/* XXX optimize by combining m_adj+M_PREPEND */
|
|
|
|
m_adj(m, sizeof(struct ether_header) - sizeof(struct llc));
|
|
|
|
llc = mtod(m, struct llc *);
|
|
|
|
llc->llc_dsap = llc->llc_ssap = LLC_SNAP_LSAP;
|
|
|
|
llc->llc_control = LLC_UI;
|
|
|
|
llc->llc_snap.org_code[0] = 0;
|
|
|
|
llc->llc_snap.org_code[1] = 0;
|
|
|
|
llc->llc_snap.org_code[2] = 0;
|
|
|
|
llc->llc_snap.ether_type = eh->ether_type;
|
|
|
|
payload = m->m_pkthdr.len; /* NB: w/o Ethernet header */
|
|
|
|
|
|
|
|
M_PREPEND(m, sizeof(struct ether_header), M_NOWAIT);
|
|
|
|
if (m == NULL) { /* XXX cannot happen */
|
|
|
|
IEEE80211_DPRINTF(vap, IEEE80211_MSG_SUPERG,
|
|
|
|
"%s: no space for ether_header\n", __func__);
|
|
|
|
vap->iv_stats.is_tx_nobuf++;
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
ETHER_HEADER_COPY(mtod(m, void *), eh);
|
|
|
|
mtod(m, struct ether_header *)->ether_type = htons(payload);
|
|
|
|
return m;
|
|
|
|
}
|
2013-08-27 14:37:13 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Complete an mbuf transmission.
|
|
|
|
*
|
|
|
|
* For now, this simply processes a completed frame after the
|
|
|
|
* driver has completed it's transmission and/or retransmission.
|
|
|
|
* It assumes the frame is an 802.11 encapsulated frame.
|
|
|
|
*
|
|
|
|
* Later on it will grow to become the exit path for a given frame
|
|
|
|
* from the driver and, depending upon how it's been encapsulated
|
|
|
|
* and already transmitted, it may end up doing A-MPDU retransmission,
|
|
|
|
* power save requeuing, etc.
|
|
|
|
*
|
|
|
|
* In order for the above to work, the driver entry point to this
|
|
|
|
* must not hold any driver locks. Thus, the driver needs to delay
|
|
|
|
* any actual mbuf completion until it can release said locks.
|
|
|
|
*
|
|
|
|
* This frees the mbuf and if the mbuf has a node reference,
|
|
|
|
* the node reference will be freed.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
ieee80211_tx_complete(struct ieee80211_node *ni, struct mbuf *m, int status)
|
|
|
|
{
|
|
|
|
|
|
|
|
if (ni != NULL) {
|
|
|
|
if (m->m_flags & M_TXCB)
|
|
|
|
ieee80211_process_callback(ni, m, status);
|
|
|
|
ieee80211_free_node(ni);
|
|
|
|
}
|
|
|
|
m_freem(m);
|
|
|
|
}
|