d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
0cfbbe3bde
freebsd-dev/sys/net80211/ieee80211_output.c

596 lines
18 KiB
C
Raw Normal View History

new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
/*-
revise copyright notices per discussion with Atsushi Onoe <onoe@sm.sony.co.jp>
2003-06-27 05:13:52 +00:00
* Copyright (c) 2001 Atsushi Onoe
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
* Copyright (c) 2002, 2003 Sam Leffler, Errno Consulting
* 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
revise copyright notices per discussion with Atsushi Onoe <onoe@sm.sony.co.jp>
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.
* 3. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
*
* Alternatively, this software may be distributed under the terms of the
* GNU General Public License ("GPL") version 2 as published by the Free
* Software Foundation.
*
revise copyright notices per discussion with Atsushi Onoe <onoe@sm.sony.co.jp>
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.
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/endian.h>
#include <sys/errno.h>
#include <sys/bus.h>
#include <sys/proc.h>
#include <sys/sysctl.h>
#include <machine/atomic.h>
#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_media.h>
#include <net/if_arp.h>
#include <net/ethernet.h>
#include <net/if_llc.h>
#include <net80211/ieee80211_var.h>
#include <net/bpf.h>
#ifdef INET
#include <netinet/in.h>
#include <netinet/if_ether.h>
#endif
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
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
* reference (and potentially free'ing up any associated storage).
*/
static int
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
ieee80211_mgmt_output(struct ifnet *ifp, struct ieee80211_node *ni,
struct mbuf *m, int type)
{
struct ieee80211com *ic = (void *)ifp;
struct ieee80211_frame *wh;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
KASSERT(ni != NULL, ("null node"));
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
ni->ni_inact = 0;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
/*
* Yech, hack alert! We want to pass the node down to the
* driver's start routine. If we don't do so then the start
* routine must immediately look it up again and that can
* cause a lock order reversal if, for example, this frame
* is being sent because the station is being timedout and
* the frame being sent is a DEAUTH message. We could stick
* this in an m_tag and tack that on to the mbuf. However
* that's rather expensive to do for every frame so instead
* we stuff it in the rcvif field since outbound frames do
* not (presently) use this.
*/
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
M_PREPEND(m, sizeof(struct ieee80211_frame), M_DONTWAIT);
if (m == NULL)
return ENOMEM;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
KASSERT(m->m_pkthdr.rcvif == NULL, ("rcvif not null"));
m->m_pkthdr.rcvif = (void *)ni;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
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_MGT | type;
wh->i_fc[1] = IEEE80211_FC1_DIR_NODS;
*(u_int16_t *)wh->i_dur = 0;
*(u_int16_t *)wh->i_seq =
htole16(ni->ni_txseq << IEEE80211_SEQ_SEQ_SHIFT);
ni->ni_txseq++;
IEEE80211_ADDR_COPY(wh->i_addr1, ni->ni_macaddr);
IEEE80211_ADDR_COPY(wh->i_addr2, ic->ic_myaddr);
IEEE80211_ADDR_COPY(wh->i_addr3, ni->ni_bssid);
if (ifp->if_flags & IFF_DEBUG) {
/* avoid to print too many frames */
if (ic->ic_opmode == IEEE80211_M_IBSS ||
#ifdef IEEE80211_DEBUG
ieee80211_debug > 1 ||
#endif
(type & IEEE80211_FC0_SUBTYPE_MASK) !=
IEEE80211_FC0_SUBTYPE_PROBE_RESP)
if_printf(ifp, "sending %s to %s on channel %u\n",
ieee80211_mgt_subtype_name[
(type & IEEE80211_FC0_SUBTYPE_MASK)
>> IEEE80211_FC0_SUBTYPE_SHIFT],
ether_sprintf(ni->ni_macaddr),
ieee80211_chan2ieee(ic, ni->ni_chan));
}
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
IF_ENQUEUE(&ic->ic_mgtq, m);
ifp->if_timer = 1;
(*ifp->if_start)(ifp);
return 0;
}
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
/*
* Encapsulate an outbound data frame. The mbuf chain is updated and
* a reference to the destination node is returned. If an error is
* encountered NULL is returned and the node reference will also be NULL.
*
* NB: The caller is responsible for free'ing a returned node reference.
* The convention is ic_bss is not reference counted; the caller must
* maintain that.
*/
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
struct mbuf *
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
ieee80211_encap(struct ifnet *ifp, struct mbuf *m, struct ieee80211_node **pni)
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
{
struct ieee80211com *ic = (void *)ifp;
struct ether_header eh;
struct ieee80211_frame *wh;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
struct ieee80211_node *ni = NULL;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
struct llc *llc;
if (m->m_len < sizeof(struct ether_header)) {
m = m_pullup(m, sizeof(struct ether_header));
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
if (m == NULL) {
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
ic->ic_stats.is_tx_nombuf++;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
goto bad;
}
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
}
memcpy(&eh, mtod(m, caddr_t), sizeof(struct ether_header));
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
if (ic->ic_opmode != IEEE80211_M_STA) {
ni = ieee80211_find_node(ic, eh.ether_dhost);
if (ni == NULL) {
/*
o add capability to indicate if device receives all management frames o use recv mgmt capability to decide if outbound frames should be discarded if no node table entry is present
2003-10-17 21:54:59 +00:00
* When not in station mode the destination
* address should always be in the node table
* if the device sends management frames to us;
* unless this is a multicast/broadcast frame.
* For devices that don't send management frames
* to the host we have to cheat; use the bss
* node instead; the card will/should clobber
* the bssid address as necessary.
*
* XXX this handles AHDEMO because all devices
* that support it don't send mgmt frames;
* but it might be better to test explicitly
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
*/
o add capability to indicate if device receives all management frames o use recv mgmt capability to decide if outbound frames should be discarded if no node table entry is present
2003-10-17 21:54:59 +00:00
if (!IEEE80211_IS_MULTICAST(eh.ether_dhost) &&
(ic->ic_caps & IEEE80211_C_RCVMGT)) {
IEEE80211_DPRINTF(("%s: no node for dst %s, "
"discard frame\n", __func__,
ether_sprintf(eh.ether_dhost)));
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
ic->ic_stats.is_tx_nonode++;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
goto bad;
}
ni = ic->ic_bss;
}
} else
ni = ic->ic_bss;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
ni->ni_inact = 0;
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;
M_PREPEND(m, sizeof(struct ieee80211_frame), M_DONTWAIT);
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
if (m == NULL) {
ic->ic_stats.is_tx_nombuf++;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
goto bad;
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
}
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
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;
*(u_int16_t *)wh->i_dur = 0;
*(u_int16_t *)wh->i_seq =
htole16(ni->ni_txseq << IEEE80211_SEQ_SEQ_SHIFT);
ni->ni_txseq++;
switch (ic->ic_opmode) {
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);
IEEE80211_ADDR_COPY(wh->i_addr3, ni->ni_bssid);
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;
add monitor mode
2003-07-21 02:49:42 +00:00
case IEEE80211_M_MONITOR:
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
goto bad;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
}
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
*pni = ni;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
return m;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
bad:
if (m != NULL)
m_freem(m);
if (ni && ni != ic->ic_bss)
ieee80211_free_node(ic, ni);
*pni = NULL;
return NULL;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
}
/*
* Add a supported rates element id to a frame.
*/
u_int8_t *
ieee80211_add_rates(u_int8_t *frm, const struct ieee80211_rateset *rs)
{
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.
*/
u_int8_t *
ieee80211_add_xrates(u_int8_t *frm, const struct ieee80211_rateset *rs)
{
/*
* 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;
}
/*
* Add an ssid elemet to a frame.
*/
static u_int8_t *
ieee80211_add_ssid(u_int8_t *frm, const u_int8_t *ssid, u_int len)
{
*frm++ = IEEE80211_ELEMID_SSID;
*frm++ = len;
memcpy(frm, ssid, len);
return frm + len;
}
static struct mbuf *
ieee80211_getmbuf(int flags, int type, u_int pktlen)
{
struct mbuf *m;
o correct logic that checks frame size to decide if a cluaster is needed o add an assertion to check the max possible packet size Noticed by: David Young <dyoung@pobox.com>
2003-09-01 02:55:09 +00:00
KASSERT(pktlen <= MCLBYTES, ("802.11 packet too large: %u", pktlen));
if (pktlen <= MHLEN)
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
MGETHDR(m, flags, type);
else
m = m_getcl(flags, type, M_PKTHDR);
return m;
}
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
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.
*/
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
int
ieee80211_send_mgmt(struct ieee80211com *ic, struct ieee80211_node *ni,
int type, int arg)
{
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
#define senderr(_x, _v) do { ic->ic_stats._v++; ret = _x; goto bad; } while (0)
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
struct ifnet *ifp = &ic->ic_if;
struct mbuf *m;
u_int8_t *frm;
enum ieee80211_phymode mode;
u_int16_t capinfo;
int ret, timer;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
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
* will remove our reference.
*/
if (ni != ic->ic_bss)
ieee80211_ref_node(ni);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
timer = 0;
switch (type) {
case IEEE80211_FC0_SUBTYPE_PROBE_REQ:
/*
* prreq frame format
* [tlv] ssid
* [tlv] supported rates
* [tlv] extended supported rates
*/
m = ieee80211_getmbuf(M_DONTWAIT, MT_DATA,
2 + ic->ic_des_esslen
+ 2 + IEEE80211_RATE_SIZE
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE));
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
m->m_data += sizeof(struct ieee80211_frame);
frm = mtod(m, u_int8_t *);
frm = ieee80211_add_ssid(frm, ic->ic_des_essid, ic->ic_des_esslen);
mode = ieee80211_chan2mode(ic, ni->ni_chan);
frm = ieee80211_add_rates(frm, &ic->ic_sup_rates[mode]);
frm = ieee80211_add_xrates(frm, &ic->ic_sup_rates[mode]);
m->m_pkthdr.len = m->m_len = frm - mtod(m, u_int8_t *);
timer = IEEE80211_TRANS_WAIT;
break;
case IEEE80211_FC0_SUBTYPE_PROBE_RESP:
/*
* probe response frame format
* [8] time stamp
* [2] beacon interval
* [2] cabability information
* [tlv] ssid
* [tlv] supported rates
include FH/DS parameters element in probe response frames
2003-10-17 22:09:20 +00:00
* [tlv] parameter set (FH/DS)
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
* [tlv] parameter set (IBSS)
* [tlv] extended supported rates
*/
m = ieee80211_getmbuf(M_DONTWAIT, MT_DATA,
8 + 2 + 2 + 2
+ 2 + ni->ni_esslen
+ 2 + IEEE80211_RATE_SIZE
include FH/DS parameters element in probe response frames
2003-10-17 22:09:20 +00:00
+ (ic->ic_phytype == IEEE80211_T_FH ? 7 : 3)
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
+ 6
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE));
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
m->m_data += sizeof(struct ieee80211_frame);
frm = mtod(m, u_int8_t *);
memset(frm, 0, 8); /* timestamp should be filled later */
frm += 8;
*(u_int16_t *)frm = htole16(ic->ic_bss->ni_intval);
frm += 2;
if (ic->ic_opmode == IEEE80211_M_IBSS)
capinfo = IEEE80211_CAPINFO_IBSS;
else
capinfo = IEEE80211_CAPINFO_ESS;
if (ic->ic_flags & IEEE80211_F_WEPON)
capinfo |= IEEE80211_CAPINFO_PRIVACY;
Honor the short preamble capability/state flag when calculating the capabilities for outbound management frames. But beware of sending this when operating on 5GHz channels; some 11a AP's reject association requests if this bit is set in the capabilities listed. Obtained from: MADWIFI (with modifications)
2003-09-14 22:34:24 +00:00
if ((ic->ic_flags & IEEE80211_F_SHPREAMBLE) &&
IEEE80211_IS_CHAN_2GHZ(ni->ni_chan))
capinfo |= IEEE80211_CAPINFO_SHORT_PREAMBLE;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
*(u_int16_t *)frm = htole16(capinfo);
frm += 2;
frm = ieee80211_add_ssid(frm, ic->ic_bss->ni_essid,
ic->ic_bss->ni_esslen);
frm = ieee80211_add_rates(frm, &ic->ic_bss->ni_rates);
include FH/DS parameters element in probe response frames
2003-10-17 22:09:20 +00:00
if (ic->ic_phytype == IEEE80211_T_FH) {
*frm++ = IEEE80211_ELEMID_FHPARMS;
*frm++ = 5;
*frm++ = ni->ni_fhdwell & 0x00ff;
*frm++ = (ni->ni_fhdwell >> 8) & 0x00ff;
*frm++ = IEEE80211_FH_CHANSET(
ieee80211_chan2ieee(ic, ni->ni_chan));
*frm++ = IEEE80211_FH_CHANPAT(
ieee80211_chan2ieee(ic, ni->ni_chan));
*frm++ = ni->ni_fhindex;
} else {
*frm++ = IEEE80211_ELEMID_DSPARMS;
*frm++ = 1;
*frm++ = ieee80211_chan2ieee(ic, ni->ni_chan);
}
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
if (ic->ic_opmode == IEEE80211_M_IBSS) {
*frm++ = IEEE80211_ELEMID_IBSSPARMS;
*frm++ = 2;
*frm++ = 0; *frm++ = 0; /* TODO: ATIM window */
} else { /* IEEE80211_M_HOSTAP */
/* TODO: TIM */
*frm++ = IEEE80211_ELEMID_TIM;
*frm++ = 4; /* length */
*frm++ = 0; /* DTIM count */
*frm++ = 1; /* DTIM period */
*frm++ = 0; /* bitmap control */
*frm++ = 0; /* Partial Virtual Bitmap (variable length) */
}
frm = ieee80211_add_xrates(frm, &ic->ic_bss->ni_rates);
m->m_pkthdr.len = m->m_len = frm - mtod(m, u_int8_t *);
break;
case IEEE80211_FC0_SUBTYPE_AUTH:
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
MH_ALIGN(m, 2 * 3);
m->m_pkthdr.len = m->m_len = 6;
frm = mtod(m, u_int8_t *);
/* TODO: shared key auth */
((u_int16_t *)frm)[0] = htole16(IEEE80211_AUTH_ALG_OPEN);
((u_int16_t *)frm)[1] = htole16(arg); /* sequence number */
((u_int16_t *)frm)[2] = 0; /* status */
if (ic->ic_opmode == IEEE80211_M_STA)
timer = IEEE80211_TRANS_WAIT;
break;
case IEEE80211_FC0_SUBTYPE_DEAUTH:
if (ifp->if_flags & IFF_DEBUG)
if_printf(ifp, "station %s deauthenticate (reason %d)\n",
ether_sprintf(ni->ni_macaddr), arg);
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
MH_ALIGN(m, 2);
m->m_pkthdr.len = m->m_len = 2;
*mtod(m, u_int16_t *) = htole16(arg); /* reason */
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
*/
m = ieee80211_getmbuf(M_DONTWAIT, MT_DATA,
sizeof(capinfo)
+ sizeof(u_int16_t)
+ IEEE80211_ADDR_LEN
+ 2 + ni->ni_esslen
+ 2 + IEEE80211_RATE_SIZE
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE));
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
m->m_data += sizeof(struct ieee80211_frame);
frm = mtod(m, u_int8_t *);
capinfo = 0;
if (ic->ic_opmode == IEEE80211_M_IBSS)
capinfo |= IEEE80211_CAPINFO_IBSS;
else /* IEEE80211_M_STA */
capinfo |= IEEE80211_CAPINFO_ESS;
if (ic->ic_flags & IEEE80211_F_WEPON)
capinfo |= IEEE80211_CAPINFO_PRIVACY;
Honor the short preamble capability/state flag when calculating the capabilities for outbound management frames. But beware of sending this when operating on 5GHz channels; some 11a AP's reject association requests if this bit is set in the capabilities listed. Obtained from: MADWIFI (with modifications)
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) &&
IEEE80211_IS_CHAN_2GHZ(ni->ni_chan))
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
capinfo |= IEEE80211_CAPINFO_SHORT_PREAMBLE;
if (ic->ic_flags & IEEE80211_F_SHSLOT)
capinfo |= IEEE80211_CAPINFO_SHORT_SLOTTIME;
*(u_int16_t *)frm = htole16(capinfo);
frm += 2;
*(u_int16_t *)frm = htole16(ic->ic_lintval);
frm += 2;
if (type == IEEE80211_FC0_SUBTYPE_REASSOC_REQ) {
IEEE80211_ADDR_COPY(frm, ic->ic_bss->ni_bssid);
frm += IEEE80211_ADDR_LEN;
}
frm = ieee80211_add_ssid(frm, ni->ni_essid, ni->ni_esslen);
frm = ieee80211_add_rates(frm, &ni->ni_rates);
frm = ieee80211_add_xrates(frm, &ni->ni_rates);
m->m_pkthdr.len = m->m_len = frm - mtod(m, u_int8_t *);
timer = IEEE80211_TRANS_WAIT;
break;
case IEEE80211_FC0_SUBTYPE_ASSOC_RESP:
case IEEE80211_FC0_SUBTYPE_REASSOC_RESP:
/*
* asreq frame format
* [2] capability information
* [2] status
* [2] association ID
* [tlv] supported rates
* [tlv] extended supported rates
*/
m = ieee80211_getmbuf(M_DONTWAIT, MT_DATA,
sizeof(capinfo)
+ sizeof(u_int16_t)
+ sizeof(u_int16_t)
+ 2 + IEEE80211_RATE_SIZE
+ 2 + (IEEE80211_RATE_MAXSIZE - IEEE80211_RATE_SIZE));
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
m->m_data += sizeof(struct ieee80211_frame);
frm = mtod(m, u_int8_t *);
capinfo = IEEE80211_CAPINFO_ESS;
if (ic->ic_flags & IEEE80211_F_WEPON)
capinfo |= IEEE80211_CAPINFO_PRIVACY;
Honor the short preamble capability/state flag when calculating the capabilities for outbound management frames. But beware of sending this when operating on 5GHz channels; some 11a AP's reject association requests if this bit is set in the capabilities listed. Obtained from: MADWIFI (with modifications)
2003-09-14 22:34:24 +00:00
if ((ic->ic_flags & IEEE80211_F_SHPREAMBLE) &&
IEEE80211_IS_CHAN_2GHZ(ni->ni_chan))
capinfo |= IEEE80211_CAPINFO_SHORT_PREAMBLE;
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
*(u_int16_t *)frm = htole16(capinfo);
frm += 2;
*(u_int16_t *)frm = htole16(arg); /* status */
frm += 2;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
if (arg == IEEE80211_STATUS_SUCCESS)
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
*(u_int16_t *)frm = htole16(ni->ni_associd);
frm += 2;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
frm = ieee80211_add_rates(frm, &ni->ni_rates);
frm = ieee80211_add_xrates(frm, &ni->ni_rates);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
m->m_pkthdr.len = m->m_len = frm - mtod(m, u_int8_t *);
break;
case IEEE80211_FC0_SUBTYPE_DISASSOC:
if (ifp->if_flags & IFF_DEBUG)
if_printf(ifp, "station %s disassociate (reason %d)\n",
ether_sprintf(ni->ni_macaddr), arg);
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL)
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(ENOMEM, is_tx_nombuf);
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
MH_ALIGN(m, 2);
m->m_pkthdr.len = m->m_len = 2;
*mtod(m, u_int16_t *) = htole16(arg); /* reason */
break;
default:
IEEE80211_DPRINTF(("%s: invalid mgmt frame type %u\n",
__func__, type));
add statistics for all failures and/or abnormal events; still need to add per-node statistics
2003-10-17 23:15:30 +00:00
senderr(EINVAL, is_tx_unknownmgt);
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
/* NOTREACHED */
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
}
ret = ieee80211_mgmt_output(ifp, ni, m, type);
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
if (ret == 0) {
if (timer)
ic->ic_mgt_timer = timer;
} else {
bad:
if (ni != ic->ic_bss) /* remove ref we added */
ieee80211_free_node(ic, ni);
}
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
return ret;
MFp4 changes to fix locking issues and correct reference count handling of station entries in hostap mode: Input path: o driver is now expected to find the node associated with the sender of a received frame; use ic_bss if none is located o driver passes the (referenced) node into ieee80211_input for use within the wlan module and is responsible for cleaning up on return o the antenna state is no longer passed up with each frame; this is now considered driver-private state and drivers are responsible for keeping it in the driver-private part of a node Output path: Revamp output path for management frames to eliminate redundant locking that causes problems and to correct reference counting bogosity that occurs when stations are timed out due to inactivity (in AP mode). On output the refcnt'd node is stashed in the pkthdr's recvif field (yech) and retrieved by the driver. This eliminates an unref/ref scenario and related node table unlock/lock due to the driver looking up the node. This is particularly important when stations are timed out as this causes a lock order reversal that can result in a deadlock. As a byproduct we also reduce the overhead for sending management frames (minimal). Additional fallout from this is a change to ieee80211_encap to return a refcn't node for tieing to the outbound frame. Node refcnts are not reclaimed until after a frame is completely processed (e.g. in the tx interrupt handler). This is especially important for timed out stations as this deref will be the final one causing the node entry to be reclaimed. Additional semi-related changes: o replace m_copym use with m_copypacket (optimization) o add assert to verify ic_bss is never free'd during normal operation o add comments explaining calling conventions by drivers for frames going in each direction o remove extraneous code that "cannot be executed" (e.g. because pointers may never be null)
2003-08-19 22:17:04 +00:00
#undef senderr
new 802.11 layer: o code reorg (relative to old netbsd-derived code) for future growth o drivers now specify available channels and rates and 802.11 layer handles almost all ifmedia actions o multi-mode support for 11a/b/g devices o 11g protocol additions (incomplete) o new element id additions (for other than 11g) o node/station table redone for proper locking and to eliminate driver incestuousness o split device flags and capabilities to reduce confusion and provide room for expansion o incomplete power management infrastructure (need to revisit) o incomplete hooks for software retry o more...
2003-06-23 16:55:01 +00:00
}
Reference in New Issue Copy Permalink