Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
b08ad7c74c
NetBSD/sys/dev/ic/wi.c

2947 lines
78 KiB
C
Raw Normal View History

If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
/* $NetBSD: wi.c,v 1.154 2004/02/10 01:08:05 dyoung Exp $ */
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
/*
* Copyright (c) 1997, 1998, 1999
* Bill Paul <wpaul@ctr.columbia.edu>. 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
* 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. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Bill Paul.
* 4. Neither the name of the author nor the names of any co-contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD
* 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.
*/
/*
* Lucent WaveLAN/IEEE 802.11 PCMCIA driver for NetBSD.
*
* Original FreeBSD driver written by Bill Paul <wpaul@ctr.columbia.edu>
* Electrical Engineering Department
* Columbia University, New York City
*/
/*
* The WaveLAN/IEEE adapter is the second generation of the WaveLAN
* from Lucent. Unlike the older cards, the new ones are programmed
* entirely via a firmware-driven controller called the Hermes.
* Unfortunately, Lucent will not release the Hermes programming manual
* without an NDA (if at all). What they do release is an API library
* called the HCF (Hardware Control Functions) which is supposed to
* do the device-specific operations of a device driver for you. The
* publically available version of the HCF library (the 'HCF Light') is
* a) extremely gross, b) lacks certain features, particularly support
* for 802.11 frames, and c) is contaminated by the GNU Public License.
*
* This driver does not use the HCF or HCF Light at all. Instead, it
* programs the Hermes controller directly, using information gleaned
* from the HCF Light code and corresponding documentation.
*
* This driver supports both the PCMCIA and ISA versions of the
* WaveLAN/IEEE cards. Note however that the ISA card isn't really
* anything of the sort: it's actually a PCMCIA bridge adapter
* that fits into an ISA slot, into which a PCMCIA WaveLAN card is
* inserted. Consequently, you need to use the pccard support for
* both the ISA and PCMCIA adapters.
*/
/*
* FreeBSD driver ported to NetBSD by Bill Sommerfeld in the back of the
* Oslo IETF plenary meeting.
*/
add/cleanup RCSID
2001-11-13 16:14:31 +03:00
#include <sys/cdefs.h>
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
__KERNEL_RCSID(0, "$NetBSD: wi.c,v 1.154 2004/02/10 01:08:05 dyoung Exp $");
add/cleanup RCSID
2001-11-13 16:14:31 +03:00
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
#define WI_HERMES_AUTOINC_WAR /* Work around data write autoinc bug. */
#define WI_HERMES_STATS_WAR /* Work around stats counter bug. */
#include "bpfilter.h"
#include <sys/param.h>
#include <sys/systm.h>
New callout mechanism with two major improvements over the old timeout()/untimeout() API: - Clients supply callout handle storage, thus eliminating problems of resource allocation. - Insertion and removal of callouts is constant time, important as this facility is used quite a lot in the kernel. The old timeout()/untimeout() API has been removed from the kernel.
2000-03-23 10:01:25 +03:00
#include <sys/callout.h>
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
#include <sys/device.h>
#include <sys/socket.h>
#include <sys/mbuf.h>
#include <sys/ioctl.h>
#include <sys/kernel.h> /* for hz */
bring over WEP mode from FreeBSD/OpenBSD
2000-02-04 10:48:29 +03:00
#include <sys/proc.h>
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
#include <net/if.h>
#include <net/if_dl.h>
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
#include <net/if_llc.h>
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
#include <net/if_media.h>
#include <net/if_ether.h>
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
#include <net80211/ieee80211_var.h>
#include <net80211/ieee80211_compat.h>
#include <net80211/ieee80211_ioctl.h>
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
#include <net80211/ieee80211_radiotap.h>
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
#include <net80211/ieee80211_rssadapt.h>
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Add shutdown hook to turn off card. Fix BPF tap code.
1999-07-15 03:07:29 +04:00
#if NBPFILTER > 0
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
#include <net/bpf.h>
#include <net/bpfdesc.h>
#endif
The back end driver doesn't need PCMCIA headers, but it does need <machine/bus.h>.
2001-06-28 02:22:12 +04:00
#include <machine/bus.h>
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
#include <dev/ic/wi_ieee.h>
#include <dev/ic/wireg.h>
#include <dev/ic/wivar.h>
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int wi_init(struct ifnet *);
static void wi_stop(struct ifnet *, int);
static void wi_start(struct ifnet *);
static int wi_reset(struct wi_softc *);
static void wi_watchdog(struct ifnet *);
static int wi_ioctl(struct ifnet *, u_long, caddr_t);
static int wi_media_change(struct ifnet *);
static void wi_media_status(struct ifnet *, struct ifmediareq *);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
static struct ieee80211_node *wi_node_alloc(struct ieee80211com *);
static void wi_node_copy(struct ieee80211com *, struct ieee80211_node *,
const struct ieee80211_node *);
static void wi_node_free(struct ieee80211com *, struct ieee80211_node *);
static void wi_raise_rate(struct ieee80211com *, struct ieee80211_rssdesc *);
static void wi_lower_rate(struct ieee80211com *, struct ieee80211_rssdesc *);
static void wi_choose_rate(struct ieee80211com *, struct ieee80211_node *,
struct ieee80211_frame *, u_int);
static void wi_rssadapt_updatestats_cb(void *, struct ieee80211_node *);
static void wi_rssadapt_updatestats(void *);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void wi_rx_intr(struct wi_softc *);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
static void wi_txalloc_intr(struct wi_softc *);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void wi_tx_intr(struct wi_softc *);
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
static void wi_tx_ex_intr(struct wi_softc *);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void wi_info_intr(struct wi_softc *);
static int wi_get_cfg(struct ifnet *, u_long, caddr_t);
static int wi_set_cfg(struct ifnet *, u_long, caddr_t);
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
static int wi_cfg_txrate(struct wi_softc *);
static int wi_write_txrate(struct wi_softc *, int);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int wi_write_wep(struct wi_softc *);
static int wi_write_multi(struct wi_softc *);
static int wi_alloc_fid(struct wi_softc *, int, int *);
static void wi_read_nicid(struct wi_softc *);
static int wi_write_ssid(struct wi_softc *, int, u_int8_t *, int);
static int wi_cmd(struct wi_softc *, int, int, int, int);
static int wi_seek_bap(struct wi_softc *, int, int);
static int wi_read_bap(struct wi_softc *, int, int, void *, int);
static int wi_write_bap(struct wi_softc *, int, int, void *, int);
wi_start: write an mbuf chain to the driver using wi_mwrite_bap, which is safer than the loop there used to be here. wi_mwrite_bap: if wi_write_bap fails, don't keep on going: this way you avoid writing garbage to the radio. First time you see an odd-length mbuf, copy the remainder of the chain to sc_txbuf and from there to the MAC. This way, you do not read an mbuf past the end of its data (occasionally you will cross a page doing that!) and you avoid expensive, excess seeks in the radio's own buffer chain. wi_rx_intr: clamp the frame length told to us by the driver to the most bytes we can fit in our mbuf cluster. I am still getting e-mails from my testers telling me how much better this makes things.
2003-01-09 11:49:39 +03:00
static int wi_mwrite_bap(struct wi_softc *, int, int, struct mbuf *, int);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int wi_read_rid(struct wi_softc *, int, void *, int *);
static int wi_write_rid(struct wi_softc *, int, void *, int);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
static int wi_newstate(struct ieee80211com *, enum ieee80211_state, int);
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
static int wi_set_tim(struct ieee80211com *, int, int);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
static int wi_scan_ap(struct wi_softc *, u_int16_t, u_int16_t);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void wi_scan_result(struct wi_softc *, int, int);
Debug code from FreeBSD: handy packet dumps when IFF_DEBUG|IFF_LINK2 are set.
2003-05-13 10:48:56 +04:00
static void wi_dump_pkt(struct wi_frame *, struct ieee80211_node *, int rssi);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static inline int
wi_write_val(struct wi_softc *sc, int rid, u_int16_t val)
{
val = htole16(val);
return wi_write_rid(sc, rid, &val, sizeof(val));
}
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
static struct timeval lasttxerror; /* time of last tx error msg */
Initialize and reset curtxeps to zero each time the error rate is exceeded. This stops the continuous spewing of error messages, making it just sporadic.
2003-05-17 20:46:03 +04:00
static int curtxeps = 0; /* current tx error msgs/sec */
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
static int wi_txerate = 0; /* tx error rate: max msgs/sec */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#ifdef WI_DEBUG
Oops. This turns off debugging again.
2003-01-09 11:52:19 +03:00
int wi_debug = 0;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#define DPRINTF(X) if (wi_debug) printf X
#define DPRINTF2(X) if (wi_debug > 1) printf X
Debug code from FreeBSD: handy packet dumps when IFF_DEBUG|IFF_LINK2 are set.
2003-05-13 10:48:56 +04:00
#define IFF_DUMPPKTS(_ifp) \
(((_ifp)->if_flags & (IFF_DEBUG|IFF_LINK2)) == (IFF_DEBUG|IFF_LINK2))
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#else
#define DPRINTF(X)
#define DPRINTF2(X)
Debug code from FreeBSD: handy packet dumps when IFF_DEBUG|IFF_LINK2 are set.
2003-05-13 10:48:56 +04:00
#define IFF_DUMPPKTS(_ifp) 0
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
#endif
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#define WI_INTRS (WI_EV_RX | WI_EV_ALLOC | WI_EV_INFO)
struct wi_card_ident
wi_card_ident[] = {
Fix some whitespace.
2002-08-11 03:16:14 +04:00
/* CARD_ID CARD_NAME FIRM_TYPE */
fix minor bugs for lucent cards. (point out from Warner Losh)
2002-04-05 04:54:51 +04:00
{ WI_NIC_LUCENT_ID, WI_NIC_LUCENT_STR, WI_LUCENT },
{ WI_NIC_SONY_ID, WI_NIC_SONY_STR, WI_LUCENT },
{ WI_NIC_LUCENT_EMB_ID, WI_NIC_LUCENT_EMB_STR, WI_LUCENT },
{ WI_NIC_EVB2_ID, WI_NIC_EVB2_STR, WI_INTERSIL },
{ WI_NIC_HWB3763_ID, WI_NIC_HWB3763_STR, WI_INTERSIL },
{ WI_NIC_HWB3163_ID, WI_NIC_HWB3163_STR, WI_INTERSIL },
{ WI_NIC_HWB3163B_ID, WI_NIC_HWB3163B_STR, WI_INTERSIL },
{ WI_NIC_EVB3_ID, WI_NIC_EVB3_STR, WI_INTERSIL },
{ WI_NIC_HWB1153_ID, WI_NIC_HWB1153_STR, WI_INTERSIL },
{ WI_NIC_P2_SST_ID, WI_NIC_P2_SST_STR, WI_INTERSIL },
{ WI_NIC_EVB2_SST_ID, WI_NIC_EVB2_SST_STR, WI_INTERSIL },
{ WI_NIC_3842_EVA_ID, WI_NIC_3842_EVA_STR, WI_INTERSIL },
using table driven for Prism2 family IDs/Names.
2002-04-04 11:06:16 +04:00
{ WI_NIC_3842_PCMCIA_AMD_ID, WI_NIC_3842_PCMCIA_STR, WI_INTERSIL },
{ WI_NIC_3842_PCMCIA_SST_ID, WI_NIC_3842_PCMCIA_STR, WI_INTERSIL },
{ WI_NIC_3842_PCMCIA_ATM_ID, WI_NIC_3842_PCMCIA_STR, WI_INTERSIL },
{ WI_NIC_3842_MINI_AMD_ID, WI_NIC_3842_MINI_STR, WI_INTERSIL },
{ WI_NIC_3842_MINI_SST_ID, WI_NIC_3842_MINI_STR, WI_INTERSIL },
{ WI_NIC_3842_MINI_ATM_ID, WI_NIC_3842_MINI_STR, WI_INTERSIL },
{ WI_NIC_3842_PCI_AMD_ID, WI_NIC_3842_PCI_STR, WI_INTERSIL },
{ WI_NIC_3842_PCI_SST_ID, WI_NIC_3842_PCI_STR, WI_INTERSIL },
{ WI_NIC_3842_PCI_ATM_ID, WI_NIC_3842_PCI_STR, WI_INTERSIL },
{ WI_NIC_P3_PCMCIA_AMD_ID, WI_NIC_P3_PCMCIA_STR, WI_INTERSIL },
{ WI_NIC_P3_PCMCIA_SST_ID, WI_NIC_P3_PCMCIA_STR, WI_INTERSIL },
{ WI_NIC_P3_MINI_AMD_ID, WI_NIC_P3_MINI_STR, WI_INTERSIL },
{ WI_NIC_P3_MINI_SST_ID, WI_NIC_P3_MINI_STR, WI_INTERSIL },
{ 0, NULL, 0 },
};
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_attach(struct wi_softc *sc)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ieee80211com *ic = &sc->sc_ic;
struct ifnet *ifp = &ic->ic_if;
Mark the available channels using the ic_channels table instead of setting bits in ic_chan_avail, which is the responsibility of ieee80211_ifattach. I cannot believe the previous code worked.
2003-11-02 04:39:22 +03:00
int chan, nrate, buflen;
u_int16_t val, chanavail;
From Steve Woodford (scw@): fix wi(4) and wiconfig(8) on big-endian machines.
2003-11-16 12:41:01 +03:00
struct {
u_int16_t nrates;
char rates[IEEE80211_RATE_SIZE];
} ratebuf;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static const u_int8_t empty_macaddr[IEEE80211_ADDR_LEN] = {
- Remove redundant assignment. - Don't disestablish shutdown hook if we can't establish it. - Noneed to manage resouces individually.
2000-03-06 13:31:27 +03:00
0x00, 0x00, 0x00, 0x00, 0x00, 0x00
};
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
int s;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
s = splnet();
Add support for Corega Wireless LAN PCC-11, and Farallon SkyLINE 11Mb. Note that Corega card works at 3.3V only, then it requires cardbus code for now.
2000-07-18 18:53:26 +04:00
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
/* Make sure interrupts are disabled. */
CSR_WRITE_2(sc, WI_INT_EN, 0);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_EVENT_ACK, ~0);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
sc->sc_invalid = 0;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
/* Reset the NIC. */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (wi_reset(sc) != 0) {
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
sc->sc_invalid = 1;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return 1;
}
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
buflen = IEEE80211_ADDR_LEN;
if (wi_read_rid(sc, WI_RID_MAC_NODE, ic->ic_myaddr, &buflen) != 0 ||
IEEE80211_ADDR_EQ(ic->ic_myaddr, empty_macaddr)) {
printf(" could not get mac address, attach failed\n");
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
fix indent
2002-02-18 17:45:56 +03:00
return 1;
on attach, check if we've got meaningful mac address or not, to detect io mapping problem. on detach, handle resource deallocations more carefully.
2000-02-13 09:17:58 +03:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
printf(" 802.11 address %s\n", ether_sprintf(ic->ic_myaddr));
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
/* Read NIC identification */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_read_nicid(sc);
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
memcpy(ifp->if_xname, sc->sc_dev.dv_xname, IFNAMSIZ);
ifp->if_softc = sc;
ifp->if_start = wi_start;
ifp->if_ioctl = wi_ioctl;
ifp->if_watchdog = wi_watchdog;
Use ether_ioctl.
2000-10-12 06:24:08 +04:00
ifp->if_init = wi_init;
ifp->if_stop = wi_stop;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_flags =
IFF_SIMPLEX | IFF_BROADCAST | IFF_MULTICAST | IFF_NOTRAILERS;
ALTQ'ify.
2000-12-14 09:27:23 +03:00
IFQ_SET_READY(&ifp->if_snd);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ic->ic_phytype = IEEE80211_T_DS;
ic->ic_opmode = IEEE80211_M_STA;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_caps = IEEE80211_C_PMGT | IEEE80211_C_AHDEMO;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ic->ic_state = IEEE80211_S_INIT;
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
ic->ic_max_aid = WI_MAX_AID;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Find available channel */
Mark the available channels using the ic_channels table instead of setting bits in ic_chan_avail, which is the responsibility of ieee80211_ifattach. I cannot believe the previous code worked.
2003-11-02 04:39:22 +03:00
buflen = sizeof(chanavail);
if (wi_read_rid(sc, WI_RID_CHANNEL_LIST, &chanavail, &buflen) != 0)
chanavail = htole16(0x1fff); /* assume 1-11 */
for (chan = 16; chan > 0; chan--) {
if (!isset((u_int8_t*)&chanavail, chan - 1))
continue;
ic->ic_ibss_chan = &ic->ic_channels[chan];
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_channels[chan].ic_freq =
ieee80211_ieee2mhz(chan, IEEE80211_CHAN_2GHZ);
ic->ic_channels[chan].ic_flags = IEEE80211_CHAN_B;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
Mark the available channels using the ic_channels table instead of setting bits in ic_chan_avail, which is the responsibility of ieee80211_ifattach. I cannot believe the previous code worked.
2003-11-02 04:39:22 +03:00
/* Find default IBSS channel */
buflen = sizeof(val);
if (wi_read_rid(sc, WI_RID_OWN_CHNL, &val, &buflen) == 0) {
chan = le16toh(val);
if (isset((u_int8_t*)&chanavail, chan - 1))
ic->ic_ibss_chan = &ic->ic_channels[chan];
}
if (ic->ic_ibss_chan == NULL)
panic("%s: no available channel\n", sc->sc_dev.dv_xname);
Improved RSSI->dBm conversion, especially for Lucent cards, for which I borrowed some conversion constants from the Linux orinoco_cs driver.
2003-07-06 11:15:55 +04:00
if (sc->sc_firmware_type == WI_LUCENT) {
sc->sc_dbm_offset = WI_LUCENT_DBM_OFFSET;
} else {
buflen = sizeof(val);
if ((sc->sc_flags & WI_FLAGS_HAS_DBMADJUST) &&
wi_read_rid(sc, WI_RID_DBM_ADJUST, &val, &buflen) == 0)
sc->sc_dbm_offset = le16toh(val);
else
sc->sc_dbm_offset = WI_PRISM_DBM_OFFSET;
Support two new RIDs. WI_RID_DBM_ADJUST: In preparation for capturing Prism packets containing received signal strength indications (RSSI), read the constant for conversion from RSSI to dBm (decibels relative to a milliwatt). WI_RID_FRAG_THRESH: Get/set the packet fragmentation threshold on Lucent/Prism.
2002-11-16 09:02:53 +03:00
}
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
/*
* Set flags based on firmware version.
*/
switch (sc->sc_firmware_type) {
case WI_LUCENT:
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_flags |= WI_FLAGS_HAS_SYSSCALE;
#ifdef WI_HERMES_AUTOINC_WAR
/* XXX: not confirmed, but never seen for recent firmware */
if (sc->sc_sta_firmware_ver < 40000) {
sc->sc_flags |= WI_FLAGS_BUG_AUTOINC;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#endif
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
/* RSS rate-adaptation is known to cause STA f/w
* 8.42.1 to lock up. STA f/w 8.70.1 and 7.28.1
* appear to work. I suspect that most versions
* will work.
*/
switch (sc->sc_sta_firmware_ver) {
case 0x084201:
sc->sc_flags &= ~WI_FLAGS_RSSADAPTSTA;
break;
case 0x087001:
case 0x072801:
default:
sc->sc_flags |= WI_FLAGS_RSSADAPTSTA;
break;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (sc->sc_sta_firmware_ver >= 60000)
sc->sc_flags |= WI_FLAGS_HAS_MOR;
Add and enable monitoring mode for Lucent cards.
2003-06-19 10:16:36 +04:00
if (sc->sc_sta_firmware_ver >= 60006) {
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_caps |= IEEE80211_C_IBSS;
ic->ic_caps |= IEEE80211_C_MONITOR;
Add and enable monitoring mode for Lucent cards.
2003-06-19 10:16:36 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_ibss_port = 1;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
break;
case WI_INTERSIL:
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
sc->sc_flags |= WI_FLAGS_RSSADAPTSTA;
Only Prism is known to support WI_RID_FRAG_THRESH.
2002-11-18 18:10:22 +03:00
sc->sc_flags |= WI_FLAGS_HAS_FRAGTHR;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_flags |= WI_FLAGS_HAS_ROAMING;
sc->sc_flags |= WI_FLAGS_HAS_SYSSCALE;
The RID for the RSSI->dBm adjustment is not supported by Prism station firmware version <= 1.1.1.
2002-12-27 10:54:35 +03:00
if (sc->sc_sta_firmware_ver > 10101)
sc->sc_flags |= WI_FLAGS_HAS_DBMADJUST;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
if (sc->sc_sta_firmware_ver >= 800) {
HostAP broke in version 1.4.2 of station firmware. Try 1.3.x, 1.4.9, or later.
2003-03-27 08:00:21 +03:00
if (sc->sc_sta_firmware_ver != 10402)
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_caps |= IEEE80211_C_HOSTAP;
ic->ic_caps |= IEEE80211_C_IBSS;
ic->ic_caps |= IEEE80211_C_MONITOR;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_ibss_port = 0;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
sc->sc_alt_retry = 2;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
break;
case WI_SYMBOL:
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
sc->sc_flags |= WI_FLAGS_RSSADAPTSTA;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_flags |= WI_FLAGS_HAS_DIVERSITY;
IBSS mode does in fact work on some earlier Symbol firmware versions, as witnessed by my home network.
2002-10-25 05:35:12 +04:00
if (sc->sc_sta_firmware_ver >= 20000)
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_caps |= IEEE80211_C_IBSS;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_ibss_port = 4;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
break;
}
"Association Faild" was added to "Link Stat" messages. ; Implement wi_scan into wi.c. forces if_wi to initiate one round of access point scan. This code was written by jrb@cs.pdx.edu, modified and bug-fixed by ichiro@netbsd.org
2002-01-21 14:28:18 +03:00
bring over WEP mode from FreeBSD/OpenBSD
2000-02-04 10:48:29 +03:00
/*
* Find out if we support WEP on this card.
*/
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
buflen = sizeof(val);
if (wi_read_rid(sc, WI_RID_WEP_AVAIL, &val, &buflen) == 0 &&
val != htole16(0))
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_caps |= IEEE80211_C_WEP;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Use WI_RID_DATA_RATES to fetch the supported rates. From OpenBSD.
2002-08-11 05:35:10 +04:00
/* Find supported rates. */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
buflen = sizeof(ratebuf);
From Steve Woodford (scw@): fix wi(4) and wiconfig(8) on big-endian machines.
2003-11-16 12:41:01 +03:00
if (wi_read_rid(sc, WI_RID_DATA_RATES, &ratebuf, &buflen) == 0) {
nrate = le16toh(ratebuf.nrates);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (nrate > IEEE80211_RATE_SIZE)
nrate = IEEE80211_RATE_SIZE;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
memcpy(ic->ic_sup_rates[IEEE80211_MODE_11B].rs_rates,
From Steve Woodford (scw@): fix wi(4) and wiconfig(8) on big-endian machines.
2003-11-16 12:41:01 +03:00
&ratebuf.rates[0], nrate);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_sup_rates[IEEE80211_MODE_11B].rs_nrates = nrate;
Make the meaning of media options consistent with other wireless drivers. mediaopt adhoc 802.11 adhoc (IBSS) mode. IBSS creation is enabled for cards which can create IBSS. mediaopt adhoc,flag0 old lucent adhoc demo mode. Note that [adhoc] media options for wi driver was used for adhoc demo mode, which is NOW CHANGED.
2002-08-21 07:26:29 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
buflen = sizeof(val);
sc->sc_max_datalen = 2304;
sc->sc_rts_thresh = 2347;
Support two new RIDs. WI_RID_DBM_ADJUST: In preparation for capturing Prism packets containing received signal strength indications (RSSI), read the constant for conversion from RSSI to dBm (decibels relative to a milliwatt). WI_RID_FRAG_THRESH: Get/set the packet fragmentation threshold on Lucent/Prism.
2002-11-16 09:02:53 +03:00
sc->sc_frag_thresh = 2346;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_system_scale = 1;
Fix to use 104bit key for Intersil/Symbol firmware: It seems that the firmware accept 104bit key only if all the keys have 104bit length. We get the length of the transmit key and use it for all other keys. Perhaps we should use software WEP for such situation.
2002-10-15 12:53:46 +04:00
sc->sc_cnfauthmode = IEEE80211_AUTH_OPEN;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_roaming_mode = 1;
Read supported rates from the card and report them. Only add media types if that rate is supported. Adapted from OpenBSD.
2002-08-11 04:00:41 +04:00
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
callout_init(&sc->sc_rssadapt_ch);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
/*
* Call MI attach routines.
*/
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if_attach(ifp);
ieee80211_ifattach(ifp);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
sc->sc_newstate = ic->ic_newstate;
ic->ic_newstate = wi_newstate;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
ic->ic_node_alloc = wi_node_alloc;
ic->ic_node_free = wi_node_free;
ic->ic_node_copy = wi_node_copy;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_set_tim = wi_set_tim;
ieee80211_media_init(ifp, wi_media_change, wi_media_status);
Still more consolidation of 802.11 media-handling, moving moving common code from awi and wi into the 802.11 framework. Inspired by Sam Leffler's patches to FreeBSD.
2003-07-07 00:01:17 +04:00
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
#if NBPFILTER > 0
bpfattach2(ifp, DLT_IEEE802_11_RADIO,
sizeof(struct ieee80211_frame) + 64, &sc->sc_drvbpf);
#endif
memset(&sc->sc_rxtapu, 0, sizeof(sc->sc_rxtapu));
sc->sc_rxtap.wr_ihdr.it_len = sizeof(sc->sc_rxtapu);
sc->sc_rxtap.wr_ihdr.it_present = WI_RX_RADIOTAP_PRESENT;
memset(&sc->sc_txtapu, 0, sizeof(sc->sc_txtapu));
sc->sc_txtap.wt_ihdr.it_len = sizeof(sc->sc_txtapu);
sc->sc_txtap.wt_ihdr.it_present = WI_TX_RADIOTAP_PRESENT;
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
/* Attach is successful. */
sc->sc_attached = 1;
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
return 0;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int
wi_detach(struct wi_softc *sc)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ifnet *ifp = &sc->sc_ic.ic_if;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
int s;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (!sc->sc_attached)
return 0;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
s = splnet();
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
sc->sc_invalid = 1;
wi_stop(ifp, 1);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Delete all remaining media. */
Still more consolidation of 802.11 media-handling, moving moving common code from awi and wi into the 802.11 framework. Inspired by Sam Leffler's patches to FreeBSD.
2003-07-07 00:01:17 +04:00
ifmedia_delete_instance(&sc->sc_ic.ic_media, IFM_INST_ANY);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ieee80211_ifdetach(ifp);
if_detach(ifp);
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return 0;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
For converge with FreeBSD:
2003-05-13 11:17:46 +04:00
#ifdef __NetBSD__
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int
wi_activate(struct device *self, enum devact act)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = (struct wi_softc *)self;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
int rv = 0, s;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
s = splnet();
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (act) {
case DVACT_ACTIVATE:
rv = EOPNOTSUPP;
break;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case DVACT_DEACTIVATE:
if_deactivate(&sc->sc_ic.ic_if);
break;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return rv;
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
void
wi_power(struct wi_softc *sc, int why)
{
struct ifnet *ifp = &sc->sc_ic.ic_if;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
int s;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
s = splnet();
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (why) {
case PWR_SUSPEND:
case PWR_STANDBY:
wi_stop(ifp, 1);
break;
case PWR_RESUME:
if (ifp->if_flags & IFF_UP) {
wi_init(ifp);
(void)wi_intr(sc);
}
break;
case PWR_SOFTSUSPEND:
case PWR_SOFTSTANDBY:
case PWR_SOFTRESUME:
break;
}
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
For converge with FreeBSD:
2003-05-13 11:17:46 +04:00
#endif /* __NetBSD__ */
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
void
wi_shutdown(struct wi_softc *sc)
{
struct ifnet *ifp = &sc->sc_ic.ic_if;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (sc->sc_attached)
wi_stop(ifp, 1);
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int
wi_intr(void *arg)
{
int i;
struct wi_softc *sc = arg;
struct ifnet *ifp = &sc->sc_ic.ic_if;
Update to wi interrupt handling: No longer require a hi-low-hi transition of a status bit before we believe it. It's a performance killer, and without it there does not seem to be a race. Sam Leffler reports that wi_intr needs to be bracketed by disable/enable interrupts through WI_INT_EN, or else performance is bad on Lucent. So now we do that.
2003-05-22 10:34:45 +04:00
u_int16_t status;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (sc->sc_enabled == 0 ||
(sc->sc_dev.dv_flags & DVF_ACTIVE) == 0 ||
(ifp->if_flags & IFF_RUNNING) == 0)
return 0;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if ((ifp->if_flags & IFF_UP) == 0) {
CSR_WRITE_2(sc, WI_INT_EN, 0);
Disable interrupts before ACKing them to avoid the obvious race.
2003-02-25 03:51:14 +03:00
CSR_WRITE_2(sc, WI_EVENT_ACK, ~0);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return 1;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Update to wi interrupt handling: No longer require a hi-low-hi transition of a status bit before we believe it. It's a performance killer, and without it there does not seem to be a race. Sam Leffler reports that wi_intr needs to be bracketed by disable/enable interrupts through WI_INT_EN, or else performance is bad on Lucent. So now we do that.
2003-05-22 10:34:45 +04:00
/* This is superfluous on Prism, but Lucent breaks if we
* do not disable interrupts.
*/
CSR_WRITE_2(sc, WI_INT_EN, 0);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* maximum 10 loops per interrupt */
for (i = 0; i < 10; i++) {
/*
* Only believe a status bit when we enter wi_intr, or when
* the bit was "off" the last time through the loop. This is
* my strategy to avoid racing the hardware/firmware if I
* can re-read the event status register more quickly than
* it is updated.
*/
Update to wi interrupt handling: No longer require a hi-low-hi transition of a status bit before we believe it. It's a performance killer, and without it there does not seem to be a race. Sam Leffler reports that wi_intr needs to be bracketed by disable/enable interrupts through WI_INT_EN, or else performance is bad on Lucent. So now we do that.
2003-05-22 10:34:45 +04:00
status = CSR_READ_2(sc, WI_EVENT_STAT);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if ((status & WI_INTRS) == 0)
break;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (status & WI_EV_RX)
wi_rx_intr(sc);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (status & WI_EV_ALLOC)
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
wi_txalloc_intr(sc);
if (status & WI_EV_TX)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_tx_intr(sc);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
if (status & WI_EV_TX_EXC)
wi_tx_ex_intr(sc);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (status & WI_EV_INFO)
wi_info_intr(sc);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
if ((ifp->if_flags & IFF_OACTIVE) == 0 &&
(sc->sc_flags & WI_FLAGS_OUTRANGE) == 0 &&
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
!IFQ_IS_EMPTY(&ifp->if_snd))
wi_start(ifp);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Update to wi interrupt handling: No longer require a hi-low-hi transition of a status bit before we believe it. It's a performance killer, and without it there does not seem to be a race. Sam Leffler reports that wi_intr needs to be bracketed by disable/enable interrupts through WI_INT_EN, or else performance is bad on Lucent. So now we do that.
2003-05-22 10:34:45 +04:00
/* re-enable interrupts */
CSR_WRITE_2(sc, WI_INT_EN, WI_INTRS);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return 1;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Factor out the initialization/reset of RSS descriptors, making wi_rssdescs_init, wi_rssdescs_reset, which I will use to receover when the firmware botches up the RSS descriptor index.
2004-02-10 03:47:41 +03:00
#define arraylen(a) (sizeof(a) / sizeof((a)[0]))
static void
wi_rssdescs_init(struct wi_rssdesc (*rssd)[WI_NTXRSS], wi_rssdescq_t *rssdfree)
{
int i;
SLIST_INIT(rssdfree);
for (i = 0; i < arraylen(*rssd); i++) {
SLIST_INSERT_HEAD(rssdfree, &(*rssd)[i], rd_next);
}
}
static void
wi_rssdescs_reset(struct ieee80211com *ic, struct wi_rssdesc (*rssd)[WI_NTXRSS],
wi_rssdescq_t *rssdfree, u_int8_t (*txpending)[IEEE80211_RATE_MAXSIZE])
{
struct ieee80211_node *ni;
int i;
for (i = 0; i < arraylen(*rssd); i++) {
ni = (*rssd)[i].rd_desc.id_node;
(*rssd)[i].rd_desc.id_node = NULL;
if (ni != NULL && (ic->ic_if.if_flags & IFF_DEBUG) != 0)
printf("%s: cleaning outstanding rssadapt "
"descriptor for %s\n",
ic->ic_if.if_xname, ether_sprintf(ni->ni_macaddr));
if (ni != NULL && ni != ic->ic_bss)
ieee80211_free_node(ic, ni);
}
memset(*txpending, 0, sizeof(*txpending));
wi_rssdescs_init(rssd, rssdfree);
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_init(struct ifnet *ifp)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
struct wi_joinreq join;
int i;
int error = 0, wasenabled;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
DPRINTF(("wi_init: enabled %d\n", sc->sc_enabled));
wasenabled = sc->sc_enabled;
if (!sc->sc_enabled) {
if ((error = (*sc->sc_enable)(sc)) != 0)
goto out;
sc->sc_enabled = 1;
} else
wi_stop(ifp, 0);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Symbol firmware cannot be initialized more than once */
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
if (sc->sc_firmware_type != WI_SYMBOL || !wasenabled)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if ((error = wi_reset(sc)) != 0)
goto out;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* common 802.11 configuration */
Set IEEE80211_F_IBSSON before test in wi_init() to set CREATE_IBSS correctly for Lucent firmware to work with IBSS mode. PR 18520
2002-10-04 02:32:37 +04:00
ic->ic_flags &= ~IEEE80211_F_IBSSON;
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
sc->sc_flags &= ~WI_FLAGS_OUTRANGE;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (ic->ic_opmode) {
case IEEE80211_M_STA:
wi_write_val(sc, WI_RID_PORTTYPE, WI_PORTTYPE_BSS);
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
case IEEE80211_M_IBSS:
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_val(sc, WI_RID_PORTTYPE, sc->sc_ibss_port);
Set IEEE80211_F_IBSSON before test in wi_init() to set CREATE_IBSS correctly for Lucent firmware to work with IBSS mode. PR 18520
2002-10-04 02:32:37 +04:00
ic->ic_flags |= IEEE80211_F_IBSSON;
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
sc->sc_syn_timer = 5;
ifp->if_timer = 1;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
case IEEE80211_M_AHDEMO:
wi_write_val(sc, WI_RID_PORTTYPE, WI_PORTTYPE_ADHOC);
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case IEEE80211_M_HOSTAP:
wi_write_val(sc, WI_RID_PORTTYPE, WI_PORTTYPE_HOSTAP);
break;
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
case IEEE80211_M_MONITOR:
Add and enable monitoring mode for Lucent cards.
2003-06-19 10:16:36 +04:00
if (sc->sc_firmware_type == WI_LUCENT)
wi_write_val(sc, WI_RID_PORTTYPE, WI_PORTTYPE_ADHOC);
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
wi_cmd(sc, WI_CMD_TEST | (WI_TEST_MONITOR << 8), 0, 0, 0);
break;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Intersil interprets this RID as joining ESS even in IBSS mode */
if (sc->sc_firmware_type == WI_LUCENT &&
(ic->ic_flags & IEEE80211_F_IBSSON) && ic->ic_des_esslen > 0)
wi_write_val(sc, WI_RID_CREATE_IBSS, 1);
else
wi_write_val(sc, WI_RID_CREATE_IBSS, 0);
wi_write_val(sc, WI_RID_MAX_SLEEP, ic->ic_lintval);
wi_write_ssid(sc, WI_RID_DESIRED_SSID, ic->ic_des_essid,
ic->ic_des_esslen);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
wi_write_val(sc, WI_RID_OWN_CHNL,
ieee80211_chan2ieee(ic, ic->ic_ibss_chan));
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_ssid(sc, WI_RID_OWN_SSID, ic->ic_des_essid, ic->ic_des_esslen);
Allow SIOCSIFADDR with AF_LINK and WI_RID_MAC_NODE (wiconfig -m) to set MAC address.
2002-10-01 07:27:02 +04:00
IEEE80211_ADDR_COPY(ic->ic_myaddr, LLADDR(ifp->if_sadl));
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_rid(sc, WI_RID_MAC_NODE, ic->ic_myaddr, IEEE80211_ADDR_LEN);
wi_write_val(sc, WI_RID_PM_ENABLED,
(ic->ic_flags & IEEE80211_F_PMGTON) ? 1 : 0);
/* not yet common 802.11 configuration */
wi_write_val(sc, WI_RID_MAX_DATALEN, sc->sc_max_datalen);
wi_write_val(sc, WI_RID_RTS_THRESH, sc->sc_rts_thresh);
Only Prism is known to support WI_RID_FRAG_THRESH.
2002-11-18 18:10:22 +03:00
if (sc->sc_flags & WI_FLAGS_HAS_FRAGTHR)
wi_write_val(sc, WI_RID_FRAG_THRESH, sc->sc_frag_thresh);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* driver specific 802.11 configuration */
if (sc->sc_flags & WI_FLAGS_HAS_SYSSCALE)
wi_write_val(sc, WI_RID_SYSTEM_SCALE, sc->sc_system_scale);
if (sc->sc_flags & WI_FLAGS_HAS_ROAMING)
wi_write_val(sc, WI_RID_ROAMING_MODE, sc->sc_roaming_mode);
if (sc->sc_flags & WI_FLAGS_HAS_MOR)
wi_write_val(sc, WI_RID_MICROWAVE_OVEN, sc->sc_microwave_oven);
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
wi_cfg_txrate(sc);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_ssid(sc, WI_RID_NODENAME, sc->sc_nodename, sc->sc_nodelen);
Do not call wi_stop() with disabled flag set from wi_init(), since it sometimes call within interrupt and deleting handler causes panic. Fix to display current tx rate as media. Clear trailer bytes in ssid fields.
2002-10-01 20:11:19 +04:00
if (ic->ic_opmode == IEEE80211_M_HOSTAP &&
sc->sc_firmware_type == WI_INTERSIL) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_val(sc, WI_RID_OWN_BEACON_INT, ic->ic_lintval);
wi_write_val(sc, WI_RID_BASIC_RATE, 0x03); /* 1, 2 */
wi_write_val(sc, WI_RID_SUPPORT_RATE, 0x0f); /* 1, 2, 5.5, 11 */
wi_write_val(sc, WI_RID_DTIM_PERIOD, 1);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Shorten the retry count on Prism APs to improve the speed of rate-adaptation.
2004-02-10 03:52:12 +03:00
if (sc->sc_firmware_type == WI_INTERSIL)
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
wi_write_val(sc, WI_RID_ALT_RETRY_COUNT, sc->sc_alt_retry);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/*
* Initialize promisc mode.
Fix spelling/grammar in a comment.
2003-11-16 12:05:53 +03:00
* Being in Host-AP mode causes a great
* deal of pain if promiscuous mode is set.
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
* Therefore we avoid confusing the firmware
* and always reset promisc mode in Host-AP
* mode. Host-AP sees all the packets anyway.
*/
if (ic->ic_opmode != IEEE80211_M_HOSTAP &&
(ifp->if_flags & IFF_PROMISC) != 0) {
wi_write_val(sc, WI_RID_PROMISC, 1);
} else {
wi_write_val(sc, WI_RID_PROMISC, 0);
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Configure WEP. */
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (ic->ic_caps & IEEE80211_C_WEP)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_wep(sc);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Set multicast filter. */
wi_write_multi(sc);
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
if (sc->sc_firmware_type != WI_SYMBOL || !wasenabled) {
sc->sc_buflen = IEEE80211_MAX_LEN + sizeof(struct wi_frame);
if (sc->sc_firmware_type == WI_SYMBOL)
sc->sc_buflen = 1585; /* XXX */
for (i = 0; i < WI_NTXBUF; i++) {
error = wi_alloc_fid(sc, sc->sc_buflen,
&sc->sc_txd[i].d_fid);
if (error) {
printf("%s: tx buffer allocation failed\n",
sc->sc_dev.dv_xname);
goto out;
}
DPRINTF2(("wi_init: txbuf %d allocated %x\n", i,
sc->sc_txd[i].d_fid));
sc->sc_txd[i].d_len = 0;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
}
initialize sc_txcur and sc_txnext at wi_init to avoid reordering after reinitialized.
2002-10-01 07:24:35 +04:00
sc->sc_txcur = sc->sc_txnext = 0;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Factor out the initialization/reset of RSS descriptors, making wi_rssdescs_init, wi_rssdescs_reset, which I will use to receover when the firmware botches up the RSS descriptor index.
2004-02-10 03:47:41 +03:00
wi_rssdescs_init(&sc->sc_rssd, &sc->sc_rssdfree);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
Misc changes to wi_stop. First, block interrupts in wi_stop. (From FreeBSD.) Fix a bug in wi_stop where ejecting from the Powerbook crashes the system. Essentially, the patch makes sure we write *no* registers when we are no longer attached. Also, for consistency w/ FreeBSD, use sc_portnum throughout wi, instead of hard-coding WI_PORT0 .
2003-05-13 10:33:40 +04:00
/* Enable desired port */
wi_cmd(sc, WI_CMD_ENABLE | sc->sc_portnum, 0, 0, 0);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_flags |= IFF_RUNNING;
ifp->if_flags &= ~IFF_OACTIVE;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_state = IEEE80211_S_INIT;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
if (ic->ic_opmode == IEEE80211_M_AHDEMO ||
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
ic->ic_opmode == IEEE80211_M_MONITOR ||
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
ic->ic_opmode == IEEE80211_M_HOSTAP)
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_new_state(ic, IEEE80211_S_RUN, -1);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Enable interrupts */
CSR_WRITE_2(sc, WI_INT_EN, WI_INTRS);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
if (!wasenabled &&
ic->ic_opmode == IEEE80211_M_HOSTAP &&
sc->sc_firmware_type == WI_INTERSIL) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* XXX: some card need to be re-enabled for hostap */
wi_cmd(sc, WI_CMD_DISABLE | WI_PORT0, 0, 0, 0);
wi_cmd(sc, WI_CMD_ENABLE | WI_PORT0, 0, 0, 0);
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (ic->ic_opmode == IEEE80211_M_STA &&
((ic->ic_flags & IEEE80211_F_DESBSSID) ||
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ic->ic_des_chan != IEEE80211_CHAN_ANYC)) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
memset(&join, 0, sizeof(join));
if (ic->ic_flags & IEEE80211_F_DESBSSID)
IEEE80211_ADDR_COPY(&join.wi_bssid, ic->ic_des_bssid);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (ic->ic_des_chan != IEEE80211_CHAN_ANYC)
join.wi_chan =
htole16(ieee80211_chan2ieee(ic, ic->ic_des_chan));
On Lucent cards, indicate that changing BSSID with SIOCS80211BSSID is impossible. Never write to the Prism-only JOIN RID.
2002-12-27 11:29:46 +03:00
/* Lucent firmware does not support the JOIN RID. */
if (sc->sc_firmware_type != WI_LUCENT)
wi_write_rid(sc, WI_RID_JOIN_REQ, &join, sizeof(join));
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
out:
if (error) {
printf("%s: interface not running\n", sc->sc_dev.dv_xname);
Do not call wi_stop() with disabled flag set from wi_init(), since it sometimes call within interrupt and deleting handler causes panic. Fix to display current tx rate as media. Clear trailer bytes in ssid fields.
2002-10-01 20:11:19 +04:00
wi_stop(ifp, 0);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
DPRINTF(("wi_init: return %d\n", error));
return error;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void
wi_stop(struct ifnet *ifp, int disable)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = ifp->if_softc;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct ieee80211com *ic = &sc->sc_ic;
Factor out the initialization/reset of RSS descriptors, making wi_rssdescs_init, wi_rssdescs_reset, which I will use to receover when the firmware botches up the RSS descriptor index.
2004-02-10 03:47:41 +03:00
int s;
Misc changes to wi_stop. First, block interrupts in wi_stop. (From FreeBSD.) Fix a bug in wi_stop where ejecting from the Powerbook crashes the system. Essentially, the patch makes sure we write *no* registers when we are no longer attached. Also, for consistency w/ FreeBSD, use sc_portnum throughout wi, instead of hard-coding WI_PORT0 .
2003-05-13 10:33:40 +04:00
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
if (!sc->sc_enabled)
return;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
s = splnet();
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
DPRINTF(("wi_stop: disable %d\n", disable));
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_new_state(ic, IEEE80211_S_INIT, -1);
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
if (!sc->sc_invalid) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_INT_EN, 0);
Misc changes to wi_stop. First, block interrupts in wi_stop. (From FreeBSD.) Fix a bug in wi_stop where ejecting from the Powerbook crashes the system. Essentially, the patch makes sure we write *no* registers when we are no longer attached. Also, for consistency w/ FreeBSD, use sc_portnum throughout wi, instead of hard-coding WI_PORT0 .
2003-05-13 10:33:40 +04:00
wi_cmd(sc, WI_CMD_DISABLE | sc->sc_portnum, 0, 0, 0);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Factor out the initialization/reset of RSS descriptors, making wi_rssdescs_init, wi_rssdescs_reset, which I will use to receover when the firmware botches up the RSS descriptor index.
2004-02-10 03:47:41 +03:00
wi_rssdescs_reset(ic, &sc->sc_rssd, &sc->sc_rssdfree,
&sc->sc_txpending);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_tx_timer = 0;
sc->sc_scan_timer = 0;
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
sc->sc_syn_timer = 0;
sc->sc_false_syns = 0;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_naps = 0;
ifp->if_flags &= ~(IFF_OACTIVE | IFF_RUNNING);
ifp->if_timer = 0;
Misc changes to wi_stop. First, block interrupts in wi_stop. (From FreeBSD.) Fix a bug in wi_stop where ejecting from the Powerbook crashes the system. Essentially, the patch makes sure we write *no* registers when we are no longer attached. Also, for consistency w/ FreeBSD, use sc_portnum throughout wi, instead of hard-coding WI_PORT0 .
2003-05-13 10:33:40 +04:00
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
if (disable) {
if (sc->sc_disable)
(*sc->sc_disable)(sc);
sc->sc_enabled = 0;
}
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
/*
* Choose a data rate for a packet len bytes long that suits the packet
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
* type and the wireless conditions.
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
*
* TBD Adapt fragmentation threshold.
*/
static void
wi_choose_rate(struct ieee80211com *ic, struct ieee80211_node *ni,
struct ieee80211_frame *wh, u_int len)
{
struct wi_softc *sc = ic->ic_if.if_softc;
struct ieee80211_rssadapt *ra;
u_int16_t (*thrs)[IEEE80211_RATE_SIZE];
struct wi_node *wn;
int flags = 0, i, rateidx = 0, s, thridx, top;
struct ieee80211_rateset *rs;
s = splnet();
wn = (void*)ni;
ra = &wn->wn_rssadapt;
rs = &ni->ni_rates;
if ((wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK) == IEEE80211_FC0_TYPE_CTL)
flags |= IEEE80211_RATE_BASIC;
for (i = 0, top = IEEE80211_RSSADAPT_BKT0;
i < IEEE80211_RSSADAPT_BKTS;
i++, top <<= IEEE80211_RSSADAPT_BKTPOWER) {
thridx = i;
if (len <= top)
break;
}
thrs = &ra->ra_rate_thresh[thridx];
if (ic->ic_fixed_rate != -1) {
if ((rs->rs_rates[ic->ic_fixed_rate] & flags) == flags) {
ni->ni_txrate = ic->ic_fixed_rate;
return;
}
flags |= IEEE80211_RATE_BASIC;
i = ic->ic_fixed_rate;
} else
i = rs->rs_nrates;
while (--i >= 0) {
rateidx = i;
if ((rs->rs_rates[i] & flags) != flags)
continue;
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
if (ic->ic_opmode != IEEE80211_M_HOSTAP &&
(sc->sc_flags & WI_FLAGS_RSSADAPTSTA) == 0)
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
break;
if ((*thrs)[i] < ra->ra_avg_rssi)
break;
}
if (ic->ic_if.if_flags & IFF_DEBUG)
printf("%s: dst %s threshold[%d, %d.%d] %d < %d\n",
ic->ic_if.if_xname, ether_sprintf(wh->i_addr1), len,
(rs->rs_rates[rateidx] & IEEE80211_RATE_VAL) / 2,
(rs->rs_rates[rateidx] & IEEE80211_RATE_VAL) * 5 % 10,
(*thrs)[rateidx], ra->ra_avg_rssi);
if (ic->ic_opmode != IEEE80211_M_HOSTAP) {
/* choose the slowest pending rate so that we don't
* accidentally send a packet on the MAC's queue
* too fast. TBD find out if the MAC labels Tx
* packets w/ rate when enqueued or dequeued.
*/
for (i = 0; i < rateidx && sc->sc_txpending[i] == 0; i++);
ni->ni_txrate = i;
} else
ni->ni_txrate = rateidx;
splx(s);
return;
}
static void
wi_raise_rate(struct ieee80211com *ic, struct ieee80211_rssdesc *id)
{
struct wi_node *wn;
if (id->id_node == NULL)
return;
wn = (void*)id->id_node;
ieee80211_rssadapt_raise_rate(ic, &wn->wn_rssadapt, id);
}
static void
wi_lower_rate(struct ieee80211com *ic, struct ieee80211_rssdesc *id)
{
struct ieee80211_node *ni;
struct wi_node *wn;
int s;
s = splnet();
if ((ni = id->id_node) == NULL) {
DPRINTF(("wi_lower_rate: missing node\n"));
goto out;
}
wn = (void *)ni;
ieee80211_rssadapt_lower_rate(ic, ni, &wn->wn_rssadapt, id);
out:
splx(s);
return;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void
wi_start(struct ifnet *ifp)
{
struct wi_softc *sc = ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
struct ieee80211_node *ni;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ieee80211_frame *wh;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
struct ieee80211_rateset *rs;
struct wi_rssdesc *rd;
struct ieee80211_rssdesc *id;
wi_start: write an mbuf chain to the driver using wi_mwrite_bap, which is safer than the loop there used to be here. wi_mwrite_bap: if wi_write_bap fails, don't keep on going: this way you avoid writing garbage to the radio. First time you see an odd-length mbuf, copy the remainder of the chain to sc_txbuf and from there to the MAC. This way, you do not read an mbuf past the end of its data (occasionally you will cross a page doing that!) and you avoid expensive, excess seeks in the radio's own buffer chain. wi_rx_intr: clamp the frame length told to us by the driver to the most bytes we can fit in our mbuf cluster. I am still getting e-mails from my testers telling me how much better this makes things.
2003-01-09 11:49:39 +03:00
struct mbuf *m0;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_frame frmhdr;
int cur, fid, off;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
Stop wi panics on card ejection in Powerbook in both the enabled and disabled states. This should fix an x86 panic somebody reported on icb.
2003-05-20 05:29:35 +04:00
if (!sc->sc_enabled || sc->sc_invalid)
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
return;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
if (sc->sc_flags & WI_FLAGS_OUTRANGE)
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
return;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
memset(&frmhdr, 0, sizeof(frmhdr));
cur = sc->sc_txnext;
for (;;) {
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ni = ic->ic_bss;
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
if (!IF_IS_EMPTY(&ic->ic_mgtq)) {
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (sc->sc_txd[cur].d_len != 0 ||
SLIST_EMPTY(&sc->sc_rssdfree)) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_flags |= IFF_OACTIVE;
break;
}
IF_DEQUEUE(&ic->ic_mgtq, m0);
m_copydata(m0, 4, ETHER_ADDR_LEN * 2,
(caddr_t)&frmhdr.wi_ehdr);
frmhdr.wi_ehdr.ether_type = 0;
wh = mtod(m0, struct ieee80211_frame *);
Get the ieee80211_node for management packets from the rcvif field.
2003-11-02 03:55:46 +03:00
ni = (struct ieee80211_node *)m0->m_pkthdr.rcvif;
m0->m_pkthdr.rcvif = NULL;
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
} else if (!IF_IS_EMPTY(&ic->ic_pwrsaveq)) {
struct llc *llc;
/*
* Should these packets be processed after the
* regular packets or before? Since they are being
* probed for, they are probably less time critical
* than other packets, but, on the other hand,
* we want the power saving nodes to go back to
* sleep as quickly as possible to save power...
*/
if (ic->ic_state != IEEE80211_S_RUN)
break;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (sc->sc_txd[cur].d_len != 0 ||
SLIST_EMPTY(&sc->sc_rssdfree)) {
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
ifp->if_flags |= IFF_OACTIVE;
break;
}
IF_DEQUEUE(&ic->ic_pwrsaveq, m0);
wh = mtod(m0, struct ieee80211_frame *);
llc = (struct llc *) (wh + 1);
m_copydata(m0, 4, ETHER_ADDR_LEN * 2,
(caddr_t)&frmhdr.wi_ehdr);
frmhdr.wi_ehdr.ether_type = llc->llc_snap.ether_type;
Fix bugs in power-saving: Use a power-saving station's ieee80211_node, which is stored in the rcvif field. Do not free a node if it is put onto the power-saving queue.
2003-11-02 03:22:49 +03:00
ni = (struct ieee80211_node *)m0->m_pkthdr.rcvif;
m0->m_pkthdr.rcvif = NULL;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
} else {
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (ic->ic_state != IEEE80211_S_RUN) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
IFQ_POLL(&ifp->if_snd, m0);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (m0 == NULL) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (sc->sc_txd[cur].d_len != 0 ||
SLIST_EMPTY(&sc->sc_rssdfree)) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_flags |= IFF_OACTIVE;
break;
}
IFQ_DEQUEUE(&ifp->if_snd, m0);
ifp->if_opackets++;
m_copydata(m0, 0, ETHER_HDR_LEN,
(caddr_t)&frmhdr.wi_ehdr);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
#if NBPFILTER > 0
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (ifp->if_bpf)
bpf_mtap(ifp->if_bpf, m0);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
#endif
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if ((m0 = ieee80211_encap(ifp, m0, &ni)) == NULL) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_oerrors++;
continue;
}
wh = mtod(m0, struct ieee80211_frame *);
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
if (ic->ic_flags & IEEE80211_F_WEPON)
wh->i_fc[1] |= IEEE80211_FC1_WEP;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (ic->ic_opmode == IEEE80211_M_HOSTAP &&
!IEEE80211_IS_MULTICAST(wh->i_addr1) &&
(wh->i_fc[0] & IEEE80211_FC0_TYPE_MASK) ==
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
IEEE80211_FC0_TYPE_DATA) {
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (ni->ni_associd == 0) {
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
m_freem(m0);
ifp->if_oerrors++;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
goto next;
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
}
if (ni->ni_pwrsave & IEEE80211_PS_SLEEP) {
ieee80211_pwrsave(ic, ni, m0);
Fix bugs in power-saving: Use a power-saving station's ieee80211_node, which is stored in the rcvif field. Do not free a node if it is put onto the power-saving queue.
2003-11-02 03:22:49 +03:00
continue; /* don't free node. */
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
}
#if NBPFILTER > 0
if (ic->ic_rawbpf)
bpf_mtap(ic->ic_rawbpf, m0);
#endif
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
frmhdr.wi_tx_ctl =
htole16(WI_ENC_TX_802_11|WI_TXCNTL_TX_EX|WI_TXCNTL_TX_OK);
Shorten the retry count on Prism APs to improve the speed of rate-adaptation.
2004-02-10 03:52:12 +03:00
if (ic->ic_opmode == IEEE80211_M_HOSTAP)
frmhdr.wi_tx_ctl |= htole16(WI_TXCNTL_ALTRTRY);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (ic->ic_opmode == IEEE80211_M_HOSTAP &&
(wh->i_fc[1] & IEEE80211_FC1_WEP)) {
if ((m0 = ieee80211_wep_crypt(ifp, m0, 1)) == NULL) {
ifp->if_oerrors++;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
goto next;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
frmhdr.wi_tx_ctl |= htole16(WI_TXCNTL_NOCRYPT);
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
wi_choose_rate(ic, ni, wh, m0->m_pkthdr.len);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#if NBPFILTER > 0
if (sc->sc_drvbpf) {
struct mbuf mb;
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
struct wi_tx_radiotap_header *tap = &sc->sc_txtap;
tap->wt_rate = ni->ni_rates.rs_rates[ni->ni_txrate];
The radiotap header is little-endian.
2004-02-10 03:32:40 +03:00
tap->wt_chan_freq =
htole16(ic->ic_bss->ni_chan->ic_freq);
tap->wt_chan_flags =
htole16(ic->ic_bss->ni_chan->ic_flags);
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
The radiotap header is little-endian.
2004-02-10 03:32:40 +03:00
/* TBD tap->wt_flags */
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
M_COPY_PKTHDR(&mb, m0);
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
mb.m_data = (caddr_t)tap;
mb.m_len = tap->wt_ihdr.it_len;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
mb.m_next = m0;
mb.m_pkthdr.len += mb.m_len;
bpf_mtap(sc->sc_drvbpf, &mb);
}
#endif
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
rs = &ni->ni_rates;
rd = SLIST_FIRST(&sc->sc_rssdfree);
id = &rd->rd_desc;
id->id_len = m0->m_pkthdr.len;
sc->sc_txd[cur].d_rate = id->id_rateidx = ni->ni_txrate;
id->id_rssi = ni->ni_rssi;
frmhdr.wi_tx_idx = rd - sc->sc_rssd;
if (ic->ic_opmode == IEEE80211_M_HOSTAP)
frmhdr.wi_tx_rate = 5 * (rs->rs_rates[ni->ni_txrate] &
IEEE80211_RATE_VAL);
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
else if (sc->sc_flags & WI_FLAGS_RSSADAPTSTA)
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
(void)wi_write_txrate(sc, rs->rs_rates[ni->ni_txrate]);
Don't take the 802.11 header off of a frame until after we've given it to the driver's BPF tap.
2003-10-20 02:00:54 +04:00
m_copydata(m0, 0, sizeof(struct ieee80211_frame),
(caddr_t)&frmhdr.wi_whdr);
m_adj(m0, sizeof(struct ieee80211_frame));
frmhdr.wi_dat_len = htole16(m0->m_pkthdr.len);
Debug code from FreeBSD: handy packet dumps when IFF_DEBUG|IFF_LINK2 are set.
2003-05-13 10:48:56 +04:00
if (IFF_DUMPPKTS(ifp))
wi_dump_pkt(&frmhdr, ni, -1);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
fid = sc->sc_txd[cur].d_fid;
off = sizeof(frmhdr);
wi_start: write an mbuf chain to the driver using wi_mwrite_bap, which is safer than the loop there used to be here. wi_mwrite_bap: if wi_write_bap fails, don't keep on going: this way you avoid writing garbage to the radio. First time you see an odd-length mbuf, copy the remainder of the chain to sc_txbuf and from there to the MAC. This way, you do not read an mbuf past the end of its data (occasionally you will cross a page doing that!) and you avoid expensive, excess seeks in the radio's own buffer chain. wi_rx_intr: clamp the frame length told to us by the driver to the most bytes we can fit in our mbuf cluster. I am still getting e-mails from my testers telling me how much better this makes things.
2003-01-09 11:49:39 +03:00
if (wi_write_bap(sc, fid, 0, &frmhdr, sizeof(frmhdr)) != 0 ||
wi_mwrite_bap(sc, fid, off, m0, m0->m_pkthdr.len) != 0) {
ifp->if_oerrors++;
m_freem(m0);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
goto next;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
m_freem(m0);
sc->sc_txd[cur].d_len = off;
if (sc->sc_txcur == cur) {
if (wi_cmd(sc, WI_CMD_TX | WI_RECLAIM, fid, 0, 0)) {
printf("%s: xmit failed\n",
sc->sc_dev.dv_xname);
sc->sc_txd[cur].d_len = 0;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
goto next;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
sc->sc_txpending[ni->ni_txrate]++;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_tx_timer = 5;
ifp->if_timer = 1;
}
sc->sc_txnext = cur = (cur + 1) % WI_NTXBUF;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
SLIST_REMOVE_HEAD(&sc->sc_rssdfree, rd_next);
Disable rate-adaptation for Lucent firmware version 8.42.1, per a bug report from Simon Burge: the firmware seems to lock up. If rate-adaptation is disabled on a Lucent card, then do all the rate-adaptation work *except* for writing the new transmit rate, since I suspect that is what locks-up the firmware.
2004-02-10 03:59:38 +03:00
id->id_node = ni;
continue;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
next:
if (ni != NULL && ni != ic->ic_bss)
ieee80211_free_node(ic, ni);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int
wi_reset(struct wi_softc *sc)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i, error;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
DPRINTF(("wi_reset\n"));
Introduced sc_reset to wi_softc. The bus-specific front-end gets to fill this.
2003-03-27 07:34:16 +03:00
if (sc->sc_reset)
(*sc->sc_reset)(sc);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = 0;
for (i = 0; i < 5; i++) {
DELAY(20*1000); /* XXX: way too long! */
if ((error = wi_cmd(sc, WI_CMD_INI, 0, 0, 0)) == 0)
break;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (error) {
printf("%s: init failed\n", sc->sc_dev.dv_xname);
return error;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_INT_EN, 0);
CSR_WRITE_2(sc, WI_EVENT_ACK, ~0);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Calibrate timer. */
wi_write_val(sc, WI_RID_TICK_TIME, 0);
return 0;
}
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Big-endian support.
2001-05-16 14:45:36 +04:00
static void
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_watchdog(struct ifnet *ifp)
{
struct wi_softc *sc = ifp->if_softc;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct ieee80211com *ic = &sc->sc_ic;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_timer = 0;
if (!sc->sc_enabled)
return;
if (sc->sc_tx_timer) {
if (--sc->sc_tx_timer == 0) {
printf("%s: device timeout\n", ifp->if_xname);
ifp->if_oerrors++;
wi_init(ifp);
return;
}
ifp->if_timer = 1;
}
if (sc->sc_scan_timer) {
if (--sc->sc_scan_timer <= WI_SCAN_WAIT - WI_SCAN_INQWAIT &&
sc->sc_firmware_type == WI_INTERSIL) {
DPRINTF(("wi_watchdog: inquire scan\n"));
wi_cmd(sc, WI_CMD_INQUIRE, WI_INFO_SCAN_RESULTS, 0, 0);
}
if (sc->sc_scan_timer)
ifp->if_timer = 1;
- Disestablish an interrupt if failed to enable the card. - If card isn't enabled, just set given information into softc. Addresses PR#9457. - Some cleanups and cosmetic changes.
2000-02-28 02:10:51 +03:00
}
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
if (sc->sc_syn_timer) {
if (--sc->sc_syn_timer == 0) {
DPRINTF2(("%s: %d false syns\n",
sc->sc_dev.dv_xname, sc->sc_false_syns));
sc->sc_false_syns = 0;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_new_state(ic, IEEE80211_S_RUN, -1);
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
sc->sc_syn_timer = 5;
}
ifp->if_timer = 1;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* TODO: rate control */
ieee80211_watchdog(ifp);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
struct ifreq *ifr = (struct ifreq *)data;
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
int s, error = 0;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
- Disestablish an interrupt if failed to enable the card. - If card isn't enabled, just set given information into softc. Addresses PR#9457. - Some cleanups and cosmetic changes.
2000-02-28 02:10:51 +03:00
if ((sc->sc_dev.dv_flags & DVF_ACTIVE) == 0)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return ENXIO;
- Disestablish an interrupt if failed to enable the card. - If card isn't enabled, just set given information into softc. Addresses PR#9457. - Some cleanups and cosmetic changes.
2000-02-28 02:10:51 +03:00
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
s = splnet();
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (cmd) {
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
case SIOCSIFFLAGS:
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
/*
* Can't do promisc and hostap at the same time. If all that's
* changing is the promisc flag, try to short-circuit a call to
* wi_init() by just setting PROMISC in the hardware.
*/
change promisc mode in case of SIOCSIFFLAGS of wi_ioctl. Fix indentation.
2001-05-08 20:42:49 +04:00
if (ifp->if_flags & IFF_UP) {
if (sc->sc_enabled) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (ic->ic_opmode != IEEE80211_M_HOSTAP &&
(ifp->if_flags & IFF_PROMISC) != 0)
wi_write_val(sc, WI_RID_PROMISC, 1);
else
wi_write_val(sc, WI_RID_PROMISC, 0);
} else
error = wi_init(ifp);
} else if (sc->sc_enabled)
wi_stop(ifp, 1);
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
break;
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
case SIOCSIFMEDIA:
case SIOCGIFMEDIA:
Still more consolidation of 802.11 media-handling, moving moving common code from awi and wi into the 802.11 framework. Inspired by Sam Leffler's patches to FreeBSD.
2003-07-07 00:01:17 +04:00
error = ifmedia_ioctl(ifp, ifr, &ic->ic_media, cmd);
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case SIOCADDMULTI:
case SIOCDELMULTI:
error = (cmd == SIOCADDMULTI) ?
ether_addmulti(ifr, &sc->sc_ic.ic_ec) :
ether_delmulti(ifr, &sc->sc_ic.ic_ec);
if (error == ENETRESET) {
if (sc->sc_enabled) {
/* do not rescan */
error = wi_write_multi(sc);
} else
error = 0;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case SIOCGIFGENERIC:
error = wi_get_cfg(ifp, cmd, data);
break;
case SIOCSIFGENERIC:
error = suser(curproc->p_ucred, &curproc->p_acflag);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
if (error)
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = wi_set_cfg(ifp, cmd, data);
if (error == ENETRESET) {
Don't attempt to gather unrequested statistics. Allow SIOCSWAVELAN with WI_RID_IFACE_STATS to request an immediate statistics update. We might in the future want to have SIOCGWAVELAN with WI_RID_IFACE_STATS return an error if a statistics gathering is in progress. Then the caller could request fresh statistics and gather them when they are done. Instead, we currently just return whatever statistics are available, which may be up to 60 seconds old.
2002-03-04 04:56:12 +03:00
if (sc->sc_enabled)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = wi_init(ifp);
else
error = 0;
change the argument of SIOCS80211NWID and SIOCG80211NWID ioctls from u_int8_t array to struct ieee80211_nwid to prepend length field. The length field is necessary because IEEE 802.11 spec doesn't prohibit even '\0' for SSID. Though the name and the value of SIOC... macro is unchanged, this change breaks binary compatibility. The only affected userland program on the tree is ifconfig(8). As Jason suggested on tech-net, it is better than live with problems since there are no releases for this ioctls yet.
2000-07-05 06:35:53 +04:00
}
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
break;
On Lucent cards, indicate that changing BSSID with SIOCS80211BSSID is impossible. Never write to the Prism-only JOIN RID.
2002-12-27 11:29:46 +03:00
case SIOCS80211BSSID:
if (sc->sc_firmware_type == WI_LUCENT) {
error = ENODEV;
break;
}
/* fall through */
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
default:
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = ieee80211_ioctl(ifp, cmd, data);
if (error == ENETRESET) {
if (sc->sc_enabled)
error = wi_init(ifp);
else
error = 0;
}
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
break;
}
IEEE80211_LOCK and WI_LOCK conceal enormous differences in locking semantics on FreeBSD and NetBSD, so I am backing them out until the macro set is enriched.
2003-05-16 05:26:17 +04:00
splx(s);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return error;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
/* TBD factor with ieee80211_media_change */
Use ether_ioctl.
2000-10-12 06:24:08 +04:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_media_change(struct ifnet *ifp)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Use ether_ioctl.
2000-10-12 06:24:08 +04:00
struct wi_softc *sc = ifp->if_softc;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ieee80211com *ic = &sc->sc_ic;
struct ifmedia_entry *ime;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
enum ieee80211_opmode newmode;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i, rate, error = 0;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Still more consolidation of 802.11 media-handling, moving moving common code from awi and wi into the 802.11 framework. Inspired by Sam Leffler's patches to FreeBSD.
2003-07-07 00:01:17 +04:00
ime = ic->ic_media.ifm_cur;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (IFM_SUBTYPE(ime->ifm_media) == IFM_AUTO) {
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
i = -1;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
} else {
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct ieee80211_rateset *rs =
&ic->ic_sup_rates[ieee80211_chan2mode(ic,
ic->ic_bss->ni_chan)];
rate = ieee80211_media2rate(ime->ifm_media);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (rate == 0)
return EINVAL;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
for (i = 0; i < rs->rs_nrates; i++) {
if ((rs->rs_rates[i] & IEEE80211_RATE_VAL) == rate)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (i == rs->rs_nrates)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return EINVAL;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
}
if (ic->ic_fixed_rate != i) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ic->ic_fixed_rate = i;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
error = ENETRESET;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if ((ime->ifm_media & IFM_IEEE80211_ADHOC) &&
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
(ime->ifm_media & IFM_FLAG0))
newmode = IEEE80211_M_AHDEMO;
else if (ime->ifm_media & IFM_IEEE80211_ADHOC)
newmode = IEEE80211_M_IBSS;
else if (ime->ifm_media & IFM_IEEE80211_HOSTAP)
newmode = IEEE80211_M_HOSTAP;
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
else if (ime->ifm_media & IFM_IEEE80211_MONITOR)
newmode = IEEE80211_M_MONITOR;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
else
newmode = IEEE80211_M_STA;
if (ic->ic_opmode != newmode) {
ic->ic_opmode = newmode;
error = ENETRESET;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
if (error == ENETRESET) {
if (sc->sc_enabled)
error = wi_init(ifp);
else
error = 0;
}
Still more consolidation of 802.11 media-handling, moving moving common code from awi and wi into the 802.11 framework. Inspired by Sam Leffler's patches to FreeBSD.
2003-07-07 00:01:17 +04:00
ifp->if_baudrate = ifmedia_baudrate(ic->ic_media.ifm_cur->ifm_media);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
return error;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void
wi_media_status(struct ifnet *ifp, struct ifmediareq *imr)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
u_int16_t val;
int rate, len;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (sc->sc_enabled == 0) {
imr->ifm_active = IFM_IEEE80211 | IFM_NONE;
imr->ifm_status = 0;
return;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
imr->ifm_status = IFM_AVALID;
imr->ifm_active = IFM_IEEE80211;
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
if (ic->ic_state == IEEE80211_S_RUN &&
(sc->sc_flags & WI_FLAGS_OUTRANGE) == 0)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
imr->ifm_status |= IFM_ACTIVE;
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
len = sizeof(val);
if (wi_read_rid(sc, WI_RID_CUR_TX_RATE, &val, &len) != 0)
rate = 0;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
else {
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
/* convert to 802.11 rate */
From Steve Woodford (scw@): fix wi(4) and wiconfig(8) on big-endian machines.
2003-11-16 12:41:01 +03:00
val = le16toh(val);
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
rate = val * 2;
if (sc->sc_firmware_type == WI_LUCENT) {
if (rate == 10)
rate = 11; /* 5.5Mbps */
} else {
if (rate == 4*2)
rate = 11; /* 5.5Mbps */
else if (rate == 8*2)
rate = 22; /* 11Mbps */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
}
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
imr->ifm_active |= ieee80211_rate2media(ic, rate, IEEE80211_MODE_11B);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (ic->ic_opmode) {
case IEEE80211_M_STA:
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
case IEEE80211_M_IBSS:
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
imr->ifm_active |= IFM_IEEE80211_ADHOC;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
break;
case IEEE80211_M_AHDEMO:
imr->ifm_active |= IFM_IEEE80211_ADHOC | IFM_FLAG0;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
case IEEE80211_M_HOSTAP:
imr->ifm_active |= IFM_IEEE80211_HOSTAP;
break;
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
case IEEE80211_M_MONITOR:
imr->ifm_active |= IFM_IEEE80211_MONITOR;
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
}
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
static struct ieee80211_node *
wi_node_alloc(struct ieee80211com *ic)
{
struct wi_node *wn =
malloc(sizeof(struct wi_node), M_DEVBUF, M_NOWAIT | M_ZERO);
return wn ? &wn->wn_node : NULL;
}
static void
wi_node_free(struct ieee80211com *ic, struct ieee80211_node *ni)
{
struct wi_softc *sc = ic->ic_if.if_softc;
int i;
for (i = 0; i < WI_NTXRSS; i++) {
if (sc->sc_rssd[i].rd_desc.id_node == ni)
sc->sc_rssd[i].rd_desc.id_node = NULL;
}
free(ni, M_DEVBUF);
}
static void
wi_node_copy(struct ieee80211com *ic, struct ieee80211_node *dst,
const struct ieee80211_node *src)
{
*(struct wi_node *)dst = *(const struct wi_node *)src;
}
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
static void
wi_sync_bssid(struct wi_softc *sc, u_int8_t new_bssid[IEEE80211_ADDR_LEN])
{
struct ieee80211com *ic = &sc->sc_ic;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct ieee80211_node *ni = ic->ic_bss;
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
struct ifnet *ifp = &ic->ic_if;
if (IEEE80211_ADDR_EQ(new_bssid, ni->ni_bssid))
return;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
DPRINTF(("wi_sync_bssid: bssid %s -> ", ether_sprintf(ni->ni_bssid)));
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
DPRINTF(("%s ?\n", ether_sprintf(new_bssid)));
/* In promiscuous mode, the BSSID field is not a reliable
* indicator of the firmware's BSSID. Damp spurious
* change-of-BSSID indications.
*/
if ((ifp->if_flags & IFF_PROMISC) != 0 &&
sc->sc_false_syns >= WI_MAX_FALSE_SYNS)
return;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_new_state(ic, IEEE80211_S_RUN, -1);
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
static __inline void
wi_rssadapt_input(struct ieee80211com *ic, struct ieee80211_node *ni,
struct ieee80211_frame *wh, int rssi)
{
struct wi_node *wn;
if (ni == NULL) {
printf("%s: null node", __func__);
return;
}
wn = (void*)ni;
ieee80211_rssadapt_input(ic, ni, &wn->wn_rssadapt, rssi);
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static void
wi_rx_intr(struct wi_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
struct ifnet *ifp = &ic->ic_if;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct ieee80211_node *ni;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_frame frmhdr;
struct mbuf *m;
struct ieee80211_frame *wh;
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
int fid, len, off, rssi;
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
u_int8_t dir;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
u_int16_t status;
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
u_int32_t rstamp;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
fid = CSR_READ_2(sc, WI_RX_FID);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* First read in the frame header */
if (wi_read_bap(sc, fid, 0, &frmhdr, sizeof(frmhdr))) {
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_RX);
ifp->if_ierrors++;
DPRINTF(("wi_rx_intr: read fid %x failed\n", fid));
return;
}
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Complete prior patch: dump packets in wi_rx_intr, too.
2003-05-13 10:51:10 +04:00
if (IFF_DUMPPKTS(ifp))
wi_dump_pkt(&frmhdr, NULL, frmhdr.wi_rx_signal);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/*
* Drop undecryptable or packets with receive errors here
*/
status = le16toh(frmhdr.wi_status);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if ((status & WI_STAT_ERRSTAT) != 0 &&
ic->ic_opmode != IEEE80211_M_MONITOR) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_RX);
ifp->if_ierrors++;
DPRINTF(("wi_rx_intr: fid %x error status %x\n", fid, status));
return;
}
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
rssi = frmhdr.wi_rx_signal;
rstamp = (le16toh(frmhdr.wi_rx_tstamp0) << 16) |
le16toh(frmhdr.wi_rx_tstamp1);
Snapshot of work-in-progress: Add AP support for Prism2-based boards. WEP for APs is not yet implemented, but without WEP, this works well enough for my laptop to associate with an AP running this code. Adapted from OpenBSD.
2002-08-11 10:13:53 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
len = le16toh(frmhdr.wi_dat_len);
off = ALIGN(sizeof(struct ieee80211_frame));
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
/* Sometimes the PRISM2.x returns bogusly large frames. Except
* in monitor mode, just throw them away.
*/
wi_start: write an mbuf chain to the driver using wi_mwrite_bap, which is safer than the loop there used to be here. wi_mwrite_bap: if wi_write_bap fails, don't keep on going: this way you avoid writing garbage to the radio. First time you see an odd-length mbuf, copy the remainder of the chain to sc_txbuf and from there to the MAC. This way, you do not read an mbuf past the end of its data (occasionally you will cross a page doing that!) and you avoid expensive, excess seeks in the radio's own buffer chain. wi_rx_intr: clamp the frame length told to us by the driver to the most bytes we can fit in our mbuf cluster. I am still getting e-mails from my testers telling me how much better this makes things.
2003-01-09 11:49:39 +03:00
if (off + len > MCLBYTES) {
Add support for Prism monitor mode. From Kevin Lahey <kml@patheticgeek.net>. This patch does NOT add monitor mode support for the Lucent radios. awi(4) was only modified for compatibility with the new mediaopt. It does NOT support monitor mode. Tested by Kevin, Daniel Carosone, and I.
2003-02-25 04:57:35 +03:00
if (ic->ic_opmode != IEEE80211_M_MONITOR) {
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_RX);
ifp->if_ierrors++;
DPRINTF(("wi_rx_intr: oversized packet\n"));
return;
} else
len = 0;
wi_start: write an mbuf chain to the driver using wi_mwrite_bap, which is safer than the loop there used to be here. wi_mwrite_bap: if wi_write_bap fails, don't keep on going: this way you avoid writing garbage to the radio. First time you see an odd-length mbuf, copy the remainder of the chain to sc_txbuf and from there to the MAC. This way, you do not read an mbuf past the end of its data (occasionally you will cross a page doing that!) and you avoid expensive, excess seeks in the radio's own buffer chain. wi_rx_intr: clamp the frame length told to us by the driver to the most bytes we can fit in our mbuf cluster. I am still getting e-mails from my testers telling me how much better this makes things.
2003-01-09 11:49:39 +03:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (m == NULL) {
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_RX);
ifp->if_ierrors++;
DPRINTF(("wi_rx_intr: MGET failed\n"));
return;
}
if (off + len > MHLEN) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_RX);
m_freem(m);
ifp->if_ierrors++;
DPRINTF(("wi_rx_intr: MCLGET failed\n"));
return;
}
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
m->m_data += off - sizeof(struct ieee80211_frame);
memcpy(m->m_data, &frmhdr.wi_whdr, sizeof(struct ieee80211_frame));
wi_read_bap(sc, fid, sizeof(frmhdr),
m->m_data + sizeof(struct ieee80211_frame), len);
m->m_pkthdr.len = m->m_len = sizeof(struct ieee80211_frame) + len;
m->m_pkthdr.rcvif = ifp;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_RX);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#if NBPFILTER > 0
if (sc->sc_drvbpf) {
struct mbuf mb;
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
struct wi_rx_radiotap_header *tap = &sc->sc_rxtap;
tap->wr_rate = frmhdr.wi_rx_rate / 5;
tap->wr_antsignal = WI_RSSI_TO_DBM(sc, frmhdr.wi_rx_signal);
tap->wr_antnoise = WI_RSSI_TO_DBM(sc, frmhdr.wi_rx_silence);
The radiotap header is little-endian.
2004-02-10 03:32:40 +03:00
tap->wr_chan_freq = htole16(ic->ic_bss->ni_chan->ic_freq);
tap->wr_chan_flags = htole16(ic->ic_bss->ni_chan->ic_flags);
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
if (frmhdr.wi_status & WI_STAT_PCF)
tap->wr_flags |= IEEE80211_RADIOTAP_F_CFP;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
M_COPY_PKTHDR(&mb, m);
Add data-link type DLT_IEEE802_11_RADIO to wi and atw. DLT_IEEE802_11_RADIO lets you monitor radio stats like received signal strength, which diversity antenna was used, channel/frequency, modulation, and data rate.
2003-11-16 12:02:42 +03:00
mb.m_data = (caddr_t)tap;
mb.m_len = tap->wr_ihdr.it_len;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
mb.m_next = m;
mb.m_pkthdr.len += mb.m_len;
bpf_mtap(sc->sc_drvbpf, &mb);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#endif
wh = mtod(m, struct ieee80211_frame *);
if (wh->i_fc[1] & IEEE80211_FC1_WEP) {
/*
* WEP is decrypted by hardware. Clear WEP bit
* header for ieee80211_input().
*/
wh->i_fc[1] &= ~IEEE80211_FC1_WEP;
}
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
/* synchronize driver's BSSID with firmware's BSSID */
dir = wh->i_fc[1] & IEEE80211_FC1_DIR_MASK;
if (ic->ic_opmode == IEEE80211_M_IBSS && dir == IEEE80211_FC1_DIR_NODS)
wi_sync_bssid(sc, wh->i_addr3);
Use ieee80211_find_rxnode to attribute Rx packets to the write ieee80211_node. This reduces code duplication. It will help us support passive scanning and rate adaptation.
2003-11-02 04:55:40 +03:00
ni = ieee80211_find_rxnode(ic, wh);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_input(ifp, m, ni, rssi, rstamp);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
wi_rssadapt_input(ic, ni, wh, rssi);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
/*
* The frame may have caused the node to be marked for
* reclamation (e.g. in response to a DEAUTH message)
* so use free_node here instead of unref_node.
*/
if (ni == ic->ic_bss)
ieee80211_unref_node(&ni);
else
ieee80211_free_node(ic, ni);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
static void
wi_tx_ex_intr(struct wi_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
struct ifnet *ifp = &ic->ic_if;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
struct ieee80211_node *ni;
struct ieee80211_rssdesc *id;
struct wi_rssdesc *rssd;
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
struct wi_frame frmhdr;
int fid;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
u_int16_t status;
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
fid = CSR_READ_2(sc, WI_TX_CMP_FID);
/* Read in the frame header */
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (wi_read_bap(sc, fid, 0, &frmhdr, sizeof(frmhdr)) != 0) {
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
printf("%s: %s read fid %x failed\n", sc->sc_dev.dv_xname,
__func__, fid);
wi_rssdescs_reset(ic, &sc->sc_rssd, &sc->sc_rssdfree,
&sc->sc_txpending);
goto out;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
}
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (frmhdr.wi_tx_idx >= WI_NTXRSS) {
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
printf("%s: %s bad idx %02x\n",
sc->sc_dev.dv_xname, __func__, frmhdr.wi_tx_idx);
wi_rssdescs_reset(ic, &sc->sc_rssd, &sc->sc_rssdfree,
&sc->sc_txpending);
goto out;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
}
status = le16toh(frmhdr.wi_status);
/*
* Spontaneous station disconnects appear as xmit
* errors. Don't announce them and/or count them
* as an output error.
*/
if (ppsratecheck(&lasttxerror, &curtxeps, wi_txerate)) {
printf("%s: tx failed", sc->sc_dev.dv_xname);
if (status & WI_TXSTAT_RET_ERR)
printf(", retry limit exceeded");
if (status & WI_TXSTAT_AGED_ERR)
printf(", max transmit lifetime exceeded");
if (status & WI_TXSTAT_DISCONNECT)
printf(", port disconnected");
if (status & WI_TXSTAT_FORM_ERR)
printf(", invalid format (data len %u src %s)",
le16toh(frmhdr.wi_dat_len),
ether_sprintf(frmhdr.wi_ehdr.ether_shost));
if (status & ~0xf)
printf(", status=0x%x", status);
printf("\n");
}
ifp->if_oerrors++;
rssd = &sc->sc_rssd[frmhdr.wi_tx_idx];
id = &rssd->rd_desc;
if ((status & WI_TXSTAT_RET_ERR) != 0)
wi_lower_rate(ic, id);
ni = id->id_node;
id->id_node = NULL;
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
if (ni == NULL) {
printf("%s: %s null node, rssdesc %02x\n",
sc->sc_dev.dv_xname, __func__, frmhdr.wi_tx_idx);
goto out;
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (sc->sc_txpending[id->id_rateidx]-- == 0) {
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
printf("%s: %s txpending[%i] wraparound", sc->sc_dev.dv_xname,
__func__, id->id_rateidx);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
sc->sc_txpending[id->id_rateidx] = 0;
}
if (ni != NULL && ni != ic->ic_bss)
ieee80211_free_node(ic, ni);
SLIST_INSERT_HEAD(&sc->sc_rssdfree, rssd, rd_next);
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
out:
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
ifp->if_flags &= ~IFF_OACTIVE;
Track TX exceptions. This will eventually enable round-robin retries (for fairness) and RTM_LOSING or some such other message to indicate a losing wireless destination. From FreeBSD.
2003-05-13 11:13:49 +04:00
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_TX_EXC);
}
Use ether_ioctl.
2000-10-12 06:24:08 +04:00
static void
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
wi_txalloc_intr(struct wi_softc *sc)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ieee80211com *ic = &sc->sc_ic;
struct ifnet *ifp = &ic->ic_if;
int fid, cur;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
fid = CSR_READ_2(sc, WI_ALLOC_FID);
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_ALLOC);
Snapshot of work-in-progress: Add AP support for Prism2-based boards. WEP for APs is not yet implemented, but without WEP, this works well enough for my laptop to associate with an AP running this code. Adapted from OpenBSD.
2002-08-11 10:13:53 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
cur = sc->sc_txcur;
if (sc->sc_txd[cur].d_fid != fid) {
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
printf("%s: bad alloc %x != %x, cur %d nxt %d\n",
sc->sc_dev.dv_xname, fid, sc->sc_txd[cur].d_fid, cur,
sc->sc_txnext);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
return;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
sc->sc_tx_timer = 0;
sc->sc_txd[cur].d_len = 0;
sc->sc_txcur = cur = (cur + 1) % WI_NTXBUF;
if (sc->sc_txd[cur].d_len == 0)
ifp->if_flags &= ~IFF_OACTIVE;
else {
if (wi_cmd(sc, WI_CMD_TX | WI_RECLAIM, sc->sc_txd[cur].d_fid,
0, 0)) {
printf("%s: xmit failed\n", sc->sc_dev.dv_xname);
sc->sc_txd[cur].d_len = 0;
} else {
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
sc->sc_txpending[sc->sc_txd[cur].d_rate]++;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_tx_timer = 5;
ifp->if_timer = 1;
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
}
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
}
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
static void
wi_tx_intr(struct wi_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
struct ifnet *ifp = &ic->ic_if;
struct ieee80211_node *ni;
struct ieee80211_rssdesc *id;
struct wi_rssdesc *rssd;
struct wi_frame frmhdr;
int fid;
fid = CSR_READ_2(sc, WI_TX_CMP_FID);
/* Read in the frame header */
if (wi_read_bap(sc, fid, 0, &frmhdr, sizeof(frmhdr)) != 0) {
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
printf("%s: %s read fid %x failed\n", sc->sc_dev.dv_xname,
__func__, fid);
wi_rssdescs_reset(ic, &sc->sc_rssd, &sc->sc_rssdfree,
&sc->sc_txpending);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
goto out;
}
if (frmhdr.wi_tx_idx >= WI_NTXRSS) {
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
printf("%s: %s bad idx %02x\n",
sc->sc_dev.dv_xname, __func__, frmhdr.wi_tx_idx);
wi_rssdescs_reset(ic, &sc->sc_rssd, &sc->sc_rssdfree,
&sc->sc_txpending);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
goto out;
}
rssd = &sc->sc_rssd[frmhdr.wi_tx_idx];
id = &rssd->rd_desc;
wi_raise_rate(ic, id);
ni = id->id_node;
id->id_node = NULL;
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
if (ni == NULL) {
printf("%s: %s null node, rssdesc %02x\n",
sc->sc_dev.dv_xname, __func__, frmhdr.wi_tx_idx);
goto out;
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (sc->sc_txpending[id->id_rateidx]-- == 0) {
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
printf("%s: %s txpending[%i] wraparound", sc->sc_dev.dv_xname,
__func__, id->id_rateidx);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
sc->sc_txpending[id->id_rateidx] = 0;
}
if (ni != NULL && ni != ic->ic_bss)
ieee80211_free_node(ic, ni);
SLIST_INSERT_HEAD(&sc->sc_rssdfree, rssd, rd_next);
out:
If the firmware returns to us an RSS descriptor index that is out of bounds, then complain and recover by freeing all of the descriptors. That will usually provoke additional complaints---see the next paragraph. If the firmware returns to us an RSS descriptor that is in-bounds but out of service (id_node == NULL), then gripe and get out of the interrupt handler. ***This will still leak RSS descriptors.*** I'm working on a better solution. Possibly I will just free all of the descriptors.
2004-02-10 04:08:05 +03:00
ifp->if_flags &= ~IFF_OACTIVE;
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_TX);
}
Big-endian support.
2001-05-16 14:45:36 +04:00
static void
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_info_intr(struct wi_softc *sc)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ieee80211com *ic = &sc->sc_ic;
struct ifnet *ifp = &ic->ic_if;
int i, fid, len, off;
u_int16_t ltbuf[2];
u_int16_t stat;
u_int32_t *ptr;
fid = CSR_READ_2(sc, WI_INFO_FID);
wi_read_bap(sc, fid, 0, ltbuf, sizeof(ltbuf));
switch (le16toh(ltbuf[1])) {
case WI_INFO_LINK_STAT:
wi_read_bap(sc, fid, sizeof(ltbuf), &stat, sizeof(stat));
DPRINTF(("wi_info_intr: LINK_STAT 0x%x\n", le16toh(stat)));
switch (le16toh(stat)) {
case CONNECTED:
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
sc->sc_flags &= ~WI_FLAGS_OUTRANGE;
Update BSSID at every CONNECTED LINK_STAT notification in IBSS mode, since it can be updated by IBSS merge. Also the lucent firmware 8.10.1 changes random generated bssid every 10 seconds in IBSS creation enable. PR 18520
2002-10-07 15:01:52 +04:00
if (ic->ic_state == IEEE80211_S_RUN &&
ic->ic_opmode != IEEE80211_M_IBSS)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
/* FALLTHROUGH */
case AP_CHANGE:
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_new_state(ic, IEEE80211_S_RUN, -1);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
Linkstatus AP_IN_RANGE doesn't mean associatted, rather it just mean trying to associate.
2002-10-01 13:48:02 +04:00
case AP_IN_RANGE:
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
sc->sc_flags &= ~WI_FLAGS_OUTRANGE;
Linkstatus AP_IN_RANGE doesn't mean associatted, rather it just mean trying to associate.
2002-10-01 13:48:02 +04:00
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case AP_OUT_OF_RANGE:
if (sc->sc_firmware_type == WI_SYMBOL &&
sc->sc_scan_timer > 0) {
if (wi_cmd(sc, WI_CMD_INQUIRE,
WI_INFO_HOST_SCAN_RESULTS, 0, 0) != 0)
sc->sc_scan_timer = 0;
break;
}
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
if (ic->ic_opmode == IEEE80211_M_STA)
sc->sc_flags |= WI_FLAGS_OUTRANGE;
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case DISCONNECTED:
case ASSOC_FAILED:
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
if (ic->ic_opmode == IEEE80211_M_STA)
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ieee80211_new_state(ic, IEEE80211_S_INIT, -1);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
}
break;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_INFO_COUNTERS:
/* some card versions have a larger stats structure */
len = min(le16toh(ltbuf[0]) - 1, sizeof(sc->sc_stats) / 4);
ptr = (u_int32_t *)&sc->sc_stats;
off = sizeof(ltbuf);
for (i = 0; i < len; i++, off += 2, ptr++) {
wi_read_bap(sc, fid, off, &stat, sizeof(stat));
From Steve Woodford (scw@): fix wi(4) and wiconfig(8) on big-endian machines.
2003-11-16 12:41:01 +03:00
stat = le16toh(stat);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#ifdef WI_HERMES_STATS_WAR
if (stat & 0xf000)
stat = ~stat;
#endif
*ptr += stat;
}
ifp->if_collisions = sc->sc_stats.wi_tx_single_retries +
sc->sc_stats.wi_tx_multi_retries +
sc->sc_stats.wi_tx_retry_limit;
break;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_INFO_SCAN_RESULTS:
case WI_INFO_HOST_SCAN_RESULTS:
wi_scan_result(sc, fid, le16toh(ltbuf[0]));
break;
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
default:
DPRINTF(("wi_info_intr: got fid %x type %x len %d\n", fid,
le16toh(ltbuf[1]), le16toh(ltbuf[0])));
break;
}
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_INFO);
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int
wi_write_multi(struct wi_softc *sc)
Blitz port of the FreeBSD wi* driver for the Lucent WaveLan IEEE PCMCIA wireless LAN. Original driver written by Bill Paul <wpaul@ctr.columbia.edu>
1999-07-15 02:24:07 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ifnet *ifp = &sc->sc_ic.ic_if;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
int n;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_mcast mlist;
struct ether_multi *enm;
struct ether_multistep estep;
Big-endian support.
2001-05-16 14:45:36 +04:00
Picky cosmetic change: remove leading space.
2003-05-31 23:38:08 +04:00
if ((ifp->if_flags & IFF_PROMISC) != 0) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
allmulti:
ifp->if_flags |= IFF_ALLMULTI;
memset(&mlist, 0, sizeof(mlist));
return wi_write_rid(sc, WI_RID_MCAST_LIST, &mlist,
sizeof(mlist));
Big-endian support.
2001-05-16 14:45:36 +04:00
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
n = 0;
ETHER_FIRST_MULTI(estep, &sc->sc_ic.ic_ec, enm);
while (enm != NULL) {
/* Punt on ranges or too many multicast addresses. */
if (!IEEE80211_ADDR_EQ(enm->enm_addrlo, enm->enm_addrhi) ||
n >= sizeof(mlist) / sizeof(mlist.wi_mcast[0]))
goto allmulti;
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
IEEE80211_ADDR_COPY(&mlist.wi_mcast[n], enm->enm_addrlo);
n++;
ETHER_NEXT_MULTI(estep, enm);
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ifp->if_flags &= ~IFF_ALLMULTI;
return wi_write_rid(sc, WI_RID_MCAST_LIST, &mlist,
IEEE80211_ADDR_LEN * n);
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
static void
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
wi_read_nicid(struct wi_softc *sc)
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_card_ident *id;
char *p;
int len;
u_int16_t ver[4];
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
getting prism2 firm ver routine to wi_get_id(). if firm ver is more than 0.8c3, promiscuous mode is off. typo WI_RID_STA_IDENEITY -> WI_RID_STA_IDENTITY
2001-05-15 13:01:27 +04:00
/* getting chip identity */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
memset(ver, 0, sizeof(ver));
len = sizeof(ver);
wi_read_rid(sc, WI_RID_CARD_ID, ver, &len);
cosmetic change. change version format: <major>.<minor> variant <num>
2001-05-16 06:36:37 +04:00
printf("%s: using ", sc->sc_dev.dv_xname);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
DPRINTF2(("wi_read_nicid: CARD_ID: %x %x %x %x\n", le16toh(ver[0]), le16toh(ver[1]), le16toh(ver[2]), le16toh(ver[3])));
using table driven for Prism2 family IDs/Names.
2002-04-04 11:06:16 +04:00
fix minor bugs for lucent cards. (point out from Warner Losh)
2002-04-05 04:54:51 +04:00
sc->sc_firmware_type = WI_NOTYPE;
using table driven for Prism2 family IDs/Names.
2002-04-04 11:06:16 +04:00
for (id = wi_card_ident; id->card_name != NULL; id++) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (le16toh(ver[0]) == id->card_id) {
using table driven for Prism2 family IDs/Names.
2002-04-04 11:06:16 +04:00
printf("%s", id->card_name);
sc->sc_firmware_type = id->firm_type;
break;
}
}
fix minor bugs for lucent cards. (point out from Warner Losh)
2002-04-05 04:54:51 +04:00
if (sc->sc_firmware_type == WI_NOTYPE) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (le16toh(ver[0]) & 0x8000) {
Fix to work Symbol firmware card, which cannot be initialized more than once. Tested with 3Com 3CRWE737A and Intel Pro/Wireless 2011.
2002-03-27 11:01:22 +03:00
printf("Unknown PRISM2 chip");
sc->sc_firmware_type = WI_INTERSIL;
} else {
printf("Unknown Lucent chip");
sc->sc_firmware_type = WI_LUCENT;
}
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
}
getting prism2 firm ver routine to wi_get_id(). if firm ver is more than 0.8c3, promiscuous mode is off. typo WI_RID_STA_IDENEITY -> WI_RID_STA_IDENTITY
2001-05-15 13:01:27 +04:00
fix minor bugs for lucent cards. (point out from Warner Losh)
2002-04-05 04:54:51 +04:00
/* get primary firmware version (Only Prism chips) */
Upps, fix typo
2002-04-05 04:59:38 +04:00
if (sc->sc_firmware_type != WI_LUCENT) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
memset(ver, 0, sizeof(ver));
len = sizeof(ver);
wi_read_rid(sc, WI_RID_PRI_IDENTITY, ver, &len);
sc->sc_pri_firmware_ver = le16toh(ver[2]) * 10000 +
le16toh(ver[3]) * 100 + le16toh(ver[1]);
fix minor bugs for lucent cards. (point out from Warner Losh)
2002-04-05 04:54:51 +04:00
}
added sc_pri_firmware_ver, and print Primary/Station firmware versions. cosmetic change.
2002-03-30 19:44:59 +03:00
the kind of "firmware" was clarified. firmware -> station firmware
2002-03-30 18:58:45 +03:00
/* get station firmware version */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
memset(ver, 0, sizeof(ver));
len = sizeof(ver);
wi_read_rid(sc, WI_RID_STA_IDENTITY, ver, &len);
sc->sc_sta_firmware_ver = le16toh(ver[2]) * 10000 +
le16toh(ver[3]) * 100 + le16toh(ver[1]);
Fix to work Symbol firmware card, which cannot be initialized more than once. Tested with 3Com 3CRWE737A and Intel Pro/Wireless 2011.
2002-03-27 11:01:22 +03:00
if (sc->sc_firmware_type == WI_INTERSIL &&
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
(sc->sc_sta_firmware_ver == 10102 ||
sc->sc_sta_firmware_ver == 20102)) {
char ident[12];
memset(ident, 0, sizeof(ident));
len = sizeof(ident);
Some fixes for Symbol card. - The version string doesn't need to start with 'V'. - Symbol firmware also support CREATE_IBSS. - But it doesn't support ROAMING_MODE nor MICROWAVE_OVEN.
2002-04-14 23:55:23 +04:00
/* value should be the format like "V2.00-11" */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (wi_read_rid(sc, WI_RID_SYMBOL_IDENTITY, ident, &len) == 0 &&
*(p = (char *)ident) >= 'A' &&
Fix to work Symbol firmware card, which cannot be initialized more than once. Tested with 3Com 3CRWE737A and Intel Pro/Wireless 2011.
2002-03-27 11:01:22 +03:00
p[2] == '.' && p[5] == '-' && p[8] == '\0') {
sc->sc_firmware_type = WI_SYMBOL;
added sc_pri_firmware_ver, and print Primary/Station firmware versions. cosmetic change.
2002-03-30 19:44:59 +03:00
sc->sc_sta_firmware_ver = (p[1] - '0') * 10000 +
Fix to work Symbol firmware card, which cannot be initialized more than once. Tested with 3Com 3CRWE737A and Intel Pro/Wireless 2011.
2002-03-27 11:01:22 +03:00
(p[3] - '0') * 1000 + (p[4] - '0') * 100 +
(p[6] - '0') * 10 + (p[7] - '0');
}
}
added sc_pri_firmware_ver, and print Primary/Station firmware versions. cosmetic change.
2002-03-30 19:44:59 +03:00
Put device name at start of line.
2002-04-11 15:21:26 +04:00
printf("\n%s: %s Firmware: ", sc->sc_dev.dv_xname,
added sc_pri_firmware_ver, and print Primary/Station firmware versions. cosmetic change.
2002-03-30 19:44:59 +03:00
sc->sc_firmware_type == WI_LUCENT ? "Lucent" :
(sc->sc_firmware_type == WI_SYMBOL ? "Symbol" : "Intersil"));
if (sc->sc_firmware_type != WI_LUCENT) /* XXX */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
printf("Primary (%u.%u.%u), ",
sc->sc_pri_firmware_ver / 10000,
added sc_pri_firmware_ver, and print Primary/Station firmware versions. cosmetic change.
2002-03-30 19:44:59 +03:00
(sc->sc_pri_firmware_ver % 10000) / 100,
sc->sc_pri_firmware_ver % 100);
printf("Station (%u.%u.%u)\n",
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_sta_firmware_ver / 10000,
(sc->sc_sta_firmware_ver % 10000) / 100,
added sc_pri_firmware_ver, and print Primary/Station firmware versions. cosmetic change.
2002-03-30 19:44:59 +03:00
sc->sc_sta_firmware_ver % 100);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
static int
wi_write_ssid(struct wi_softc *sc, int rid, u_int8_t *buf, int buflen)
{
struct wi_ssid ssid;
getting prism2 firm ver routine to wi_get_id(). if firm ver is more than 0.8c3, promiscuous mode is off. typo WI_RID_STA_IDENEITY -> WI_RID_STA_IDENTITY
2001-05-15 13:01:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (buflen > IEEE80211_NWID_LEN)
return ENOBUFS;
Do not call wi_stop() with disabled flag set from wi_init(), since it sometimes call within interrupt and deleting handler causes panic. Fix to display current tx rate as media. Clear trailer bytes in ssid fields.
2002-10-01 20:11:19 +04:00
memset(&ssid, 0, sizeof(ssid));
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ssid.wi_len = htole16(buflen);
memcpy(ssid.wi_ssid, buf, buflen);
Do not call wi_stop() with disabled flag set from wi_init(), since it sometimes call within interrupt and deleting handler causes panic. Fix to display current tx rate as media. Clear trailer bytes in ssid fields.
2002-10-01 20:11:19 +04:00
return wi_write_rid(sc, rid, &ssid, sizeof(ssid));
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int
wi_get_cfg(struct ifnet *ifp, u_long cmd, caddr_t data)
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
struct ifreq *ifr = (struct ifreq *)data;
struct wi_req wreq;
int len, n, error;
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = copyin(ifr->ifr_data, &wreq, sizeof(wreq));
if (error)
return error;
len = (wreq.wi_len - 1) * 2;
if (len < sizeof(u_int16_t))
return ENOSPC;
if (len > sizeof(wreq.wi_val))
len = sizeof(wreq.wi_val);
- Remove redundant assignment. - Don't disestablish shutdown hook if we can't establish it. - Noneed to manage resouces individually.
2000-03-06 13:31:27 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (wreq.wi_type) {
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_IFACE_STATS:
memcpy(wreq.wi_val, &sc->sc_stats, sizeof(sc->sc_stats));
if (len < sizeof(sc->sc_stats))
error = ENOSPC;
else
len = sizeof(sc->sc_stats);
break;
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_ENCRYPTION:
case WI_RID_TX_CRYPT_KEY:
case WI_RID_DEFLT_CRYPT_KEYS:
case WI_RID_TX_RATE:
In wi_get_cfg, factor the code for getting the tx rate with the IEEE 802.11 subsystem. In wi_write_txrate, document the clever code for computing WI_RID_TX_RATE from a bit rate.
2002-11-16 09:25:05 +03:00
return ieee80211_cfgget(ifp, cmd, data);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_MICROWAVE_OVEN:
if (sc->sc_enabled && (sc->sc_flags & WI_FLAGS_HAS_MOR)) {
error = wi_read_rid(sc, wreq.wi_type, wreq.wi_val,
&len);
break;
}
wreq.wi_val[0] = htole16(sc->sc_microwave_oven);
len = sizeof(u_int16_t);
break;
The RID for the RSSI->dBm adjustment is not supported by Prism station firmware version <= 1.1.1.
2002-12-27 10:54:35 +03:00
case WI_RID_DBM_ADJUST:
if (sc->sc_enabled && (sc->sc_flags & WI_FLAGS_HAS_DBMADJUST)) {
error = wi_read_rid(sc, wreq.wi_type, wreq.wi_val,
&len);
break;
}
Improved RSSI->dBm conversion, especially for Lucent cards, for which I borrowed some conversion constants from the Linux orinoco_cs driver.
2003-07-06 11:15:55 +04:00
wreq.wi_val[0] = htole16(sc->sc_dbm_offset);
The RID for the RSSI->dBm adjustment is not supported by Prism station firmware version <= 1.1.1.
2002-12-27 10:54:35 +03:00
len = sizeof(u_int16_t);
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_ROAMING_MODE:
if (sc->sc_enabled && (sc->sc_flags & WI_FLAGS_HAS_ROAMING)) {
error = wi_read_rid(sc, wreq.wi_type, wreq.wi_val,
&len);
break;
}
wreq.wi_val[0] = htole16(sc->sc_roaming_mode);
len = sizeof(u_int16_t);
break;
case WI_RID_SYSTEM_SCALE:
if (sc->sc_enabled && (sc->sc_flags & WI_FLAGS_HAS_SYSSCALE)) {
error = wi_read_rid(sc, wreq.wi_type, wreq.wi_val,
&len);
break;
}
wreq.wi_val[0] = htole16(sc->sc_system_scale);
len = sizeof(u_int16_t);
break;
Only Prism is known to support WI_RID_FRAG_THRESH.
2002-11-18 18:10:22 +03:00
case WI_RID_FRAG_THRESH:
if (sc->sc_enabled && (sc->sc_flags & WI_FLAGS_HAS_FRAGTHR)) {
error = wi_read_rid(sc, wreq.wi_type, wreq.wi_val,
&len);
break;
}
wreq.wi_val[0] = htole16(sc->sc_frag_thresh);
len = sizeof(u_int16_t);
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_READ_APS:
if (ic->ic_opmode == IEEE80211_M_HOSTAP)
return ieee80211_cfgget(ifp, cmd, data);
if (sc->sc_scan_timer > 0) {
error = EINPROGRESS;
break;
}
n = sc->sc_naps;
if (len < sizeof(n)) {
error = ENOSPC;
break;
}
if (len < sizeof(n) + sizeof(struct wi_apinfo) * n)
n = (len - sizeof(n)) / sizeof(struct wi_apinfo);
len = sizeof(n) + sizeof(struct wi_apinfo) * n;
memcpy(wreq.wi_val, &n, sizeof(n));
memcpy((caddr_t)wreq.wi_val + sizeof(n), sc->sc_aps,
sizeof(struct wi_apinfo) * n);
break;
default:
if (sc->sc_enabled) {
error = wi_read_rid(sc, wreq.wi_type, wreq.wi_val,
&len);
break;
}
switch (wreq.wi_type) {
case WI_RID_MAX_DATALEN:
wreq.wi_val[0] = htole16(sc->sc_max_datalen);
len = sizeof(u_int16_t);
break;
Following implementation for other RIDs, read the driver's notion of the fragmentation threshold instead of the hardware's notion.
2003-02-25 03:47:11 +03:00
case WI_RID_FRAG_THRESH:
wreq.wi_val[0] = htole16(sc->sc_frag_thresh);
len = sizeof(u_int16_t);
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_RTS_THRESH:
wreq.wi_val[0] = htole16(sc->sc_rts_thresh);
len = sizeof(u_int16_t);
break;
case WI_RID_CNFAUTHMODE:
wreq.wi_val[0] = htole16(sc->sc_cnfauthmode);
len = sizeof(u_int16_t);
break;
case WI_RID_NODENAME:
if (len < sc->sc_nodelen + sizeof(u_int16_t)) {
error = ENOSPC;
break;
}
len = sc->sc_nodelen + sizeof(u_int16_t);
wreq.wi_val[0] = htole16((sc->sc_nodelen + 1) / 2);
memcpy(&wreq.wi_val[1], sc->sc_nodename,
sc->sc_nodelen);
break;
default:
return ieee80211_cfgget(ifp, cmd, data);
}
break;
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (error)
return error;
wreq.wi_len = (len + 1) / 2 + 1;
return copyout(&wreq, ifr->ifr_data, (wreq.wi_len + 1) * 2);
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
}
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int
wi_set_cfg(struct ifnet *ifp, u_long cmd, caddr_t data)
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_softc *sc = ifp->if_softc;
struct ieee80211com *ic = &sc->sc_ic;
struct ifreq *ifr = (struct ifreq *)data;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct ieee80211_rateset *rs = &ic->ic_sup_rates[IEEE80211_MODE_11B];
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct wi_req wreq;
struct mbuf *m;
int i, len, error;
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = copyin(ifr->ifr_data, &wreq, sizeof(wreq));
if (error)
return error;
len = (wreq.wi_len - 1) * 2;
switch (wreq.wi_type) {
The RID for the RSSI->dBm adjustment is not supported by Prism station firmware version <= 1.1.1.
2002-12-27 10:54:35 +03:00
case WI_RID_DBM_ADJUST:
return ENODEV;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_NODENAME:
if (le16toh(wreq.wi_val[0]) * 2 > len ||
le16toh(wreq.wi_val[0]) > sizeof(sc->sc_nodename)) {
error = ENOSPC;
break;
}
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
if (sc->sc_enabled) {
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = wi_write_rid(sc, wreq.wi_type, wreq.wi_val,
len);
if (error)
break;
}
sc->sc_nodelen = le16toh(wreq.wi_val[0]) * 2;
memcpy(sc->sc_nodename, &wreq.wi_val[1], sc->sc_nodelen);
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_MICROWAVE_OVEN:
case WI_RID_ROAMING_MODE:
case WI_RID_SYSTEM_SCALE:
Only Prism is known to support WI_RID_FRAG_THRESH.
2002-11-18 18:10:22 +03:00
case WI_RID_FRAG_THRESH:
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (wreq.wi_type == WI_RID_MICROWAVE_OVEN &&
(sc->sc_flags & WI_FLAGS_HAS_MOR) == 0)
break;
if (wreq.wi_type == WI_RID_ROAMING_MODE &&
(sc->sc_flags & WI_FLAGS_HAS_ROAMING) == 0)
break;
if (wreq.wi_type == WI_RID_SYSTEM_SCALE &&
(sc->sc_flags & WI_FLAGS_HAS_SYSSCALE) == 0)
break;
Only Prism is known to support WI_RID_FRAG_THRESH.
2002-11-18 18:10:22 +03:00
if (wreq.wi_type == WI_RID_FRAG_THRESH &&
(sc->sc_flags & WI_FLAGS_HAS_FRAGTHR) == 0)
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* FALLTHROUGH */
case WI_RID_RTS_THRESH:
case WI_RID_CNFAUTHMODE:
case WI_RID_MAX_DATALEN:
if (sc->sc_enabled) {
error = wi_write_rid(sc, wreq.wi_type, wreq.wi_val,
sizeof(u_int16_t));
if (error)
break;
}
switch (wreq.wi_type) {
Support two new RIDs. WI_RID_DBM_ADJUST: In preparation for capturing Prism packets containing received signal strength indications (RSSI), read the constant for conversion from RSSI to dBm (decibels relative to a milliwatt). WI_RID_FRAG_THRESH: Get/set the packet fragmentation threshold on Lucent/Prism.
2002-11-16 09:02:53 +03:00
case WI_RID_FRAG_THRESH:
sc->sc_frag_thresh = le16toh(wreq.wi_val[0]);
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_RTS_THRESH:
sc->sc_rts_thresh = le16toh(wreq.wi_val[0]);
break;
case WI_RID_MICROWAVE_OVEN:
sc->sc_microwave_oven = le16toh(wreq.wi_val[0]);
break;
case WI_RID_ROAMING_MODE:
sc->sc_roaming_mode = le16toh(wreq.wi_val[0]);
break;
case WI_RID_SYSTEM_SCALE:
sc->sc_system_scale = le16toh(wreq.wi_val[0]);
break;
case WI_RID_CNFAUTHMODE:
sc->sc_cnfauthmode = le16toh(wreq.wi_val[0]);
break;
case WI_RID_MAX_DATALEN:
sc->sc_max_datalen = le16toh(wreq.wi_val[0]);
break;
}
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_TX_RATE:
switch (le16toh(wreq.wi_val[0])) {
case 3:
ic->ic_fixed_rate = -1;
break;
default:
for (i = 0; i < IEEE80211_RATE_SIZE; i++) {
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if ((rs->rs_rates[i] & IEEE80211_RATE_VAL)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/ 2 == le16toh(wreq.wi_val[0]))
break;
}
if (i == IEEE80211_RATE_SIZE)
return EINVAL;
ic->ic_fixed_rate = i;
It enabled it to control roaming mode and authentication mode. Roaming mode can change value into 'firm mode' and disable. Authentication mode can change into 'Open System authentication' and 'Shared Key Authentication' with Prism2 chip. wi_get_id() was introduced in order that chip might judge automatically whether it is Prism2 chip. Therefore, a pp_prism2 entry in "struct wi_pcmcia_product" became unnecessary.
2001-05-15 08:14:06 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (sc->sc_enabled)
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
error = wi_cfg_txrate(sc);
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_SCAN_APS:
if (sc->sc_enabled && ic->ic_opmode != IEEE80211_M_HOSTAP)
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
error = wi_scan_ap(sc, 0x3fff, 0x000f);
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_RID_MGMT_XMIT:
if (!sc->sc_enabled) {
error = ENETDOWN;
break;
}
if (ic->ic_mgtq.ifq_len > 5) {
error = EAGAIN;
break;
}
/* XXX wi_len looks in u_int8_t, not in u_int16_t */
m = m_devget((char *)&wreq.wi_val, wreq.wi_len, 0, ifp, NULL);
if (m == NULL) {
error = ENOMEM;
break;
}
IF_ENQUEUE(&ic->ic_mgtq, m);
break;
old lucent adhoc-demo mode (adhoc,flag0 or wiconfig -p 3) wasn't handled correctly. To avoid massive extra code in each driver, now if_iee80211subr.c also handles non-standard old lucent adhoc-demo mode. This also fixes PR 14227.
2002-09-30 19:48:41 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
default:
if (sc->sc_enabled) {
error = wi_write_rid(sc, wreq.wi_type, wreq.wi_val,
len);
if (error)
break;
}
error = ieee80211_cfgset(ifp, cmd, data);
divided into PCMCIA attachment and common code of wi. since if_wi_ieee.h does not depend for pcmcia, sys/dev/pcmcia/if_wi_ieee.h was moved to sys/dev/ic/wi_ieee.h .
2001-05-06 07:26:38 +04:00
break;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return error;
add wi_detach and friends so that we can hot-swap wi card.
2000-02-12 19:08:04 +03:00
}
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
/* Rate is 0 for hardware auto-select, otherwise rate is
* 2, 4, 11, or 22 (units of 500Kbps).
*/
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
static int
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
wi_write_txrate(struct wi_softc *sc, int rate)
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
{
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
u_int16_t hwrate;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
rate = (rate & IEEE80211_RATE_VAL) / 2;
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* rate: 0, 1, 2, 5, 11 */
switch (sc->sc_firmware_type) {
case WI_LUCENT:
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
switch (rate) {
case 0:
hwrate = 3; /* auto */
break;
case 5:
hwrate = 4;
break;
case 11:
hwrate = 5;
break;
default:
hwrate = rate;
break;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
default:
In wi_get_cfg, factor the code for getting the tx rate with the IEEE 802.11 subsystem. In wi_write_txrate, document the clever code for computing WI_RID_TX_RATE from a bit rate.
2002-11-16 09:25:05 +03:00
/* Choose a bit according to this table.
*
* bit | data rate
* ----+-------------------
* 0 | 1Mbps
* 1 | 2Mbps
* 2 | 5.5Mbps
* 3 | 11Mbps
*/
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
for (i = 8; i > 0; i >>= 1) {
if (rate >= i)
break;
}
if (i == 0)
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
hwrate = 0xf; /* auto */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
else
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
hwrate = i;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
break;
}
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
if (sc->sc_tx_rate == hwrate)
return 0;
Bug fix: in wi_cfg_txrate, do not divide the 802.11 rate by 2 and take off the flags before giving it to wi_write_txrate. Also add a useful debug message to wi_write_txrate.
2003-11-02 02:57:05 +03:00
if (sc->sc_if.if_flags & IFF_DEBUG)
printf("%s: tx rate %d -> %d (%d)\n", __func__, sc->sc_tx_rate,
hwrate, rate);
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
sc->sc_tx_rate = hwrate;
return wi_write_val(sc, WI_RID_TX_RATE, sc->sc_tx_rate);
}
static int
wi_cfg_txrate(struct wi_softc *sc)
{
struct ieee80211com *ic = &sc->sc_ic;
struct ieee80211_rateset *rs;
int rate;
rs = &ic->ic_sup_rates[IEEE80211_MODE_11B];
sc->sc_tx_rate = 0; /* force write to RID */
if (ic->ic_fixed_rate < 0)
rate = 0; /* auto */
else
Bug fix: in wi_cfg_txrate, do not divide the 802.11 rate by 2 and take off the flags before giving it to wi_write_txrate. Also add a useful debug message to wi_write_txrate.
2003-11-02 02:57:05 +03:00
rate = rs->rs_rates[ic->ic_fixed_rate];
Prepare for wi rate adaptation. Add wi_write_txrate, which only writes a hardware register when the data rate selection actually changes. Add wi_cfg_txrate, which writes the data rate selection regardless. Name some fields in the wi Tx buffer which we will use for rate adaptation. Name the Prism "Alternate Retry Count" RID, WI_RID_ALT_RETRY_COUNT.
2003-10-16 14:57:35 +04:00
return wi_write_txrate(sc, rate);
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
}
Sync internal state and ifmedia state. Addresses PR#10424.
2000-07-31 07:25:11 +04:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_wep(struct wi_softc *sc)
Sync internal state and ifmedia state. Addresses PR#10424.
2000-07-31 07:25:11 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
struct ieee80211com *ic = &sc->sc_ic;
int error = 0;
int i, keylen;
u_int16_t val;
struct wi_key wkey[IEEE80211_WEP_NKID];
Sync internal state and ifmedia state. Addresses PR#10424.
2000-07-31 07:25:11 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (sc->sc_firmware_type) {
case WI_LUCENT:
val = (ic->ic_flags & IEEE80211_F_WEPON) ? 1 : 0;
error = wi_write_val(sc, WI_RID_ENCRYPTION, val);
if (error)
break;
error = wi_write_val(sc, WI_RID_TX_CRYPT_KEY, ic->ic_wep_txkey);
if (error)
break;
memset(wkey, 0, sizeof(wkey));
for (i = 0; i < IEEE80211_WEP_NKID; i++) {
keylen = ic->ic_nw_keys[i].wk_len;
wkey[i].wi_keylen = htole16(keylen);
memcpy(wkey[i].wi_keydat, ic->ic_nw_keys[i].wk_key,
keylen);
}
error = wi_write_rid(sc, WI_RID_DEFLT_CRYPT_KEYS,
wkey, sizeof(wkey));
Sync internal state and ifmedia state. Addresses PR#10424.
2000-07-31 07:25:11 +04:00
break;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_INTERSIL:
case WI_SYMBOL:
if (ic->ic_flags & IEEE80211_F_WEPON) {
/*
* ONLY HWB3163 EVAL-CARD Firmware version
* less than 0.8 variant2
*
* If promiscuous mode disable, Prism2 chip
* does not work with WEP .
* It is under investigation for details.
netbsd.org -> NetBSD.org This was the last commit of this kind to src/sys, which is now totally "NetBSD.org clean". Thanks for the patiance, and sorry for all the commits.
2003-12-04 16:57:30 +03:00
* (ichiro@NetBSD.org)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
*/
if (sc->sc_firmware_type == WI_INTERSIL &&
sc->sc_sta_firmware_ver < 802 ) {
/* firm ver < 0.8 variant 2 */
wi_write_val(sc, WI_RID_PROMISC, 1);
}
wi_write_val(sc, WI_RID_CNFAUTHMODE,
sc->sc_cnfauthmode);
val = PRIVACY_INVOKED | EXCLUDE_UNENCRYPTED;
Fix to use 104bit key for Intersil/Symbol firmware: It seems that the firmware accept 104bit key only if all the keys have 104bit length. We get the length of the transmit key and use it for all other keys. Perhaps we should use software WEP for such situation.
2002-10-15 12:53:46 +04:00
/*
* Encryption firmware has a bug for HostAP mode.
*/
if (sc->sc_firmware_type == WI_INTERSIL &&
ic->ic_opmode == IEEE80211_M_HOSTAP)
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
val |= HOST_ENCRYPT;
} else {
Fix to use 104bit key for Intersil/Symbol firmware: It seems that the firmware accept 104bit key only if all the keys have 104bit length. We get the length of the transmit key and use it for all other keys. Perhaps we should use software WEP for such situation.
2002-10-15 12:53:46 +04:00
wi_write_val(sc, WI_RID_CNFAUTHMODE,
IEEE80211_AUTH_OPEN);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
val = HOST_ENCRYPT | HOST_DECRYPT;
}
error = wi_write_val(sc, WI_RID_P2_ENCRYPTION, val);
if (error)
break;
error = wi_write_val(sc, WI_RID_P2_TX_CRYPT_KEY,
ic->ic_wep_txkey);
if (error)
break;
Fix to use 104bit key for Intersil/Symbol firmware: It seems that the firmware accept 104bit key only if all the keys have 104bit length. We get the length of the transmit key and use it for all other keys. Perhaps we should use software WEP for such situation.
2002-10-15 12:53:46 +04:00
/*
* It seems that the firmware accept 104bit key only if
* all the keys have 104bit length. We get the length of
* the transmit key and use it for all other keys.
* Perhaps we should use software WEP for such situation.
*/
keylen = ic->ic_nw_keys[ic->ic_wep_txkey].wk_len;
if (keylen > IEEE80211_WEP_KEYLEN)
keylen = 13; /* 104bit keys */
else
keylen = IEEE80211_WEP_KEYLEN;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
for (i = 0; i < IEEE80211_WEP_NKID; i++) {
error = wi_write_rid(sc, WI_RID_P2_CRYPT_KEY0 + i,
ic->ic_nw_keys[i].wk_key, keylen);
if (error)
break;
}
Sync internal state and ifmedia state. Addresses PR#10424.
2000-07-31 07:25:11 +04:00
break;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return error;
Sync internal state and ifmedia state. Addresses PR#10424.
2000-07-31 07:25:11 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Must be called at proper protection level! */
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_cmd(struct wi_softc *sc, int cmd, int val0, int val1, int val2)
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i, status;
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* wait for the busy bit to clear */
For uniformity between FreeBSD & NetBSD, adopt FreeBSD's more concise wi_cmd.
2003-05-13 10:15:47 +04:00
for (i = 500; i > 0; i--) { /* 5s */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if ((CSR_READ_2(sc, WI_COMMAND) & WI_CMD_BUSY) == 0)
break;
For uniformity between FreeBSD & NetBSD, adopt FreeBSD's more concise wi_cmd.
2003-05-13 10:15:47 +04:00
DELAY(10*1000); /* 10 m sec */
* Add "wi_flags" to the softc to indicate capabilities of the card. Fill in this info based on firmware type and version. Replace "wi_has_wep" with a flag. * For cards which support IBSS, add "ibss" has a media option. * For cards which support CREATE_IBSS, add "ibss-master" as a media option. * Use the "ibss" and "ibss-master" media options to determine if we should go into IBSS mode and/or create the IBSS. * Internally convert between the generic WI_PORTTYPE_IBSS and the value the firmware wants to use for IBSS. * When setting the IBSS name, if a desired-SSID is set, and we're an ibss-master, write the desired-SSID into the own-SSID slot. This ensures that "ifconfig wi0 nwid FOO" will do the expected thing. * Only set the roaming mode if the card supports roaming. From OpenBSD.
2002-08-11 05:30:28 +04:00
}
For uniformity between FreeBSD & NetBSD, adopt FreeBSD's more concise wi_cmd.
2003-05-13 10:15:47 +04:00
if (i == 0) {
printf("%s: wi_cmd: busy bit won't clear.\n",
sc->sc_dev.dv_xname);
return(ETIMEDOUT);
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_PARAM0, val0);
CSR_WRITE_2(sc, WI_PARAM1, val1);
CSR_WRITE_2(sc, WI_PARAM2, val2);
CSR_WRITE_2(sc, WI_COMMAND, cmd);
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (cmd == WI_CMD_INI) {
/* XXX: should sleep here. */
DELAY(100*1000);
}
/* wait for the cmd completed bit */
for (i = 0; i < WI_TIMEOUT; i++) {
if (CSR_READ_2(sc, WI_EVENT_STAT) & WI_EV_CMD)
break;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
DELAY(WI_DELAY);
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
status = CSR_READ_2(sc, WI_STATUS);
/* Ack the command */
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_CMD);
Set ifp->if_baudrate as appropriate.
2000-03-07 00:02:37 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (i == WI_TIMEOUT) {
printf("%s: command timed out, cmd=0x%x, arg=0x%x\n",
sc->sc_dev.dv_xname, cmd, val0);
return ETIMEDOUT;
}
if (status & WI_STAT_CMD_RESULT) {
printf("%s: command failed, cmd=0x%x, arg=0x%x\n",
sc->sc_dev.dv_xname, cmd, val0);
return EIO;
}
return 0;
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int
wi_seek_bap(struct wi_softc *sc, int id, int off)
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i, status;
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
CSR_WRITE_2(sc, WI_SEL0, id);
CSR_WRITE_2(sc, WI_OFF0, off);
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
for (i = 0; ; i++) {
status = CSR_READ_2(sc, WI_OFF0);
if ((status & WI_OFF_BUSY) == 0)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (i == WI_TIMEOUT) {
printf("%s: timeout in wi_seek to %x/%x\n",
sc->sc_dev.dv_xname, id, off);
sc->sc_bap_off = WI_OFF_ERR; /* invalidate */
return ETIMEDOUT;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
DELAY(1);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (status & WI_OFF_ERR) {
printf("%s: failed in wi_seek to %x/%x\n",
sc->sc_dev.dv_xname, id, off);
sc->sc_bap_off = WI_OFF_ERR; /* invalidate */
return EIO;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_bap_id = id;
sc->sc_bap_off = off;
return 0;
Add ifmedia support so that dhcp happy. Also, add nwid support.
2000-03-02 08:00:47 +03:00
}
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_read_bap(struct wi_softc *sc, int id, int off, void *buf, int buflen)
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int error, cnt;
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (buflen == 0)
return 0;
if (id != sc->sc_bap_id || off != sc->sc_bap_off) {
if ((error = wi_seek_bap(sc, id, off)) != 0)
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
return error;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
cnt = (buflen + 1) / 2;
CSR_READ_MULTI_STREAM_2(sc, WI_DATA0, (u_int16_t *)buf, cnt);
sc->sc_bap_off += cnt * 2;
return 0;
}
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
static int
wi_write_bap(struct wi_softc *sc, int id, int off, void *buf, int buflen)
{
int error, cnt;
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (buflen == 0)
return 0;
#ifdef WI_HERMES_AUTOINC_WAR
again:
#endif
if (id != sc->sc_bap_id || off != sc->sc_bap_off) {
if ((error = wi_seek_bap(sc, id, off)) != 0)
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
return error;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
cnt = (buflen + 1) / 2;
CSR_WRITE_MULTI_STREAM_2(sc, WI_DATA0, (u_int16_t *)buf, cnt);
sc->sc_bap_off += cnt * 2;
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
#ifdef WI_HERMES_AUTOINC_WAR
/*
* According to the comments in the HCF Light code, there is a bug
* in the Hermes (or possibly in certain Hermes firmware revisions)
* where the chip's internal autoincrement counter gets thrown off
* during data writes: the autoincrement is missed, causing one
* data word to be overwritten and subsequent words to be written to
* the wrong memory locations. The end result is that we could end
* up transmitting bogus frames without realizing it. The workaround
* for this is to write a couple of extra guard words after the end
* of the transfer, then attempt to read then back. If we fail to
* locate the guard words where we expect them, we preform the
* transfer over again.
*/
if ((sc->sc_flags & WI_FLAGS_BUG_AUTOINC) && (id & 0xf000) == 0) {
CSR_WRITE_2(sc, WI_DATA0, 0x1234);
CSR_WRITE_2(sc, WI_DATA0, 0x5678);
wi_seek_bap(sc, id, sc->sc_bap_off);
sc->sc_bap_off = WI_OFF_ERR; /* invalidate */
if (CSR_READ_2(sc, WI_DATA0) != 0x1234 ||
CSR_READ_2(sc, WI_DATA0) != 0x5678) {
printf("%s: detect auto increment bug, try again\n",
sc->sc_dev.dv_xname);
goto again;
}
}
#endif
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
return 0;
}
wi_start: write an mbuf chain to the driver using wi_mwrite_bap, which is safer than the loop there used to be here. wi_mwrite_bap: if wi_write_bap fails, don't keep on going: this way you avoid writing garbage to the radio. First time you see an odd-length mbuf, copy the remainder of the chain to sc_txbuf and from there to the MAC. This way, you do not read an mbuf past the end of its data (occasionally you will cross a page doing that!) and you avoid expensive, excess seeks in the radio's own buffer chain. wi_rx_intr: clamp the frame length told to us by the driver to the most bytes we can fit in our mbuf cluster. I am still getting e-mails from my testers telling me how much better this makes things.
2003-01-09 11:49:39 +03:00
static int
wi_mwrite_bap(struct wi_softc *sc, int id, int off, struct mbuf *m0, int totlen)
{
int error, len;
struct mbuf *m;
for (m = m0; m != NULL && totlen > 0; m = m->m_next) {
if (m->m_len == 0)
continue;
len = min(m->m_len, totlen);
if (((u_long)m->m_data) % 2 != 0 || len % 2 != 0) {
m_copydata(m, 0, totlen, (caddr_t)&sc->sc_txbuf);
return wi_write_bap(sc, id, off, (caddr_t)&sc->sc_txbuf,
totlen);
}
if ((error = wi_write_bap(sc, id, off, m->m_data, len)) != 0)
return error;
off += m->m_len;
totlen -= len;
}
return 0;
}
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_alloc_fid(struct wi_softc *sc, int len, int *idp)
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i;
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (wi_cmd(sc, WI_CMD_ALLOC_MEM, len, 0, 0)) {
printf("%s: failed to allocate %d bytes on NIC\n",
sc->sc_dev.dv_xname, len);
return ENOMEM;
}
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
for (i = 0; i < WI_TIMEOUT; i++) {
if (CSR_READ_2(sc, WI_EVENT_STAT) & WI_EV_ALLOC)
break;
if (i == WI_TIMEOUT) {
printf("%s: timeout in alloc\n", sc->sc_dev.dv_xname);
return ETIMEDOUT;
}
DELAY(1);
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
*idp = CSR_READ_2(sc, WI_ALLOC_FID);
CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_ALLOC);
add support for SIOCS80211NWKEY and SIOCG80211NWKEY.
2000-07-21 08:48:55 +04:00
return 0;
}
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_read_rid(struct wi_softc *sc, int rid, void *buf, int *buflenp)
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int error, len;
u_int16_t ltbuf[2];
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Tell the NIC to enter record read mode. */
error = wi_cmd(sc, WI_CMD_ACCESS | WI_ACCESS_READ, rid, 0, 0);
if (error)
return error;
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = wi_read_bap(sc, rid, 0, ltbuf, sizeof(ltbuf));
if (error)
return error;
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (le16toh(ltbuf[1]) != rid) {
printf("%s: record read mismatch, rid=%x, got=%x\n",
sc->sc_dev.dv_xname, rid, le16toh(ltbuf[1]));
return EIO;
}
len = (le16toh(ltbuf[0]) - 1) * 2; /* already got rid */
if (*buflenp < len) {
printf("%s: record buffer is too small, "
"rid=%x, size=%d, len=%d\n",
sc->sc_dev.dv_xname, rid, *buflenp, len);
return ENOSPC;
}
*buflenp = len;
return wi_read_bap(sc, rid, sizeof(ltbuf), buf, len);
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
}
static int
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
wi_write_rid(struct wi_softc *sc, int rid, void *buf, int buflen)
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int error;
u_int16_t ltbuf[2];
ltbuf[0] = htole16((buflen + 1) / 2 + 1); /* includes rid */
ltbuf[1] = htole16(rid);
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = wi_write_bap(sc, rid, 0, ltbuf, sizeof(ltbuf));
if (error)
return error;
error = wi_write_bap(sc, rid, sizeof(ltbuf), buf, buflen);
if (error)
return error;
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
return wi_cmd(sc, WI_CMD_ACCESS | WI_ACCESS_WRITE, rid, 0, 0);
Support SIOC{G,S}80211POWER.
2000-12-12 07:04:29 +03:00
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
static void
wi_rssadapt_updatestats_cb(void *arg, struct ieee80211_node *ni)
{
struct wi_node *wn = (void*)ni;
ieee80211_rssadapt_updatestats(&wn->wn_rssadapt);
}
static void
wi_rssadapt_updatestats(void *arg)
{
struct wi_softc *sc = arg;
struct ieee80211com *ic = &sc->sc_ic;
ieee80211_iterate_nodes(ic, wi_rssadapt_updatestats_cb, arg);
if (ic->ic_opmode != IEEE80211_M_MONITOR &&
ic->ic_state == IEEE80211_S_RUN)
callout_reset(&sc->sc_rssadapt_ch, hz / 10,
wi_rssadapt_updatestats, arg);
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
static int
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
wi_newstate(struct ieee80211com *ic, enum ieee80211_state nstate, int arg)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
{
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
struct wi_softc *sc = ic->ic_softc;
struct ieee80211_node *ni = ic->ic_bss;
int buflen;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
u_int16_t val;
struct wi_ssid ssid;
In wi_newstate(), read the BSSID into a temporary buffer that has guaranteed alignment. Fixes a really obscure failure mode on Alpha, and panics on other machines.
2003-10-25 03:58:22 +04:00
struct wi_macaddr bssid, old_bssid;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
enum ieee80211_state ostate;
#ifdef WI_DEBUG
static const char *stname[] =
{ "INIT", "SCAN", "AUTH", "ASSOC", "RUN" };
#endif /* WI_DEBUG */
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ostate = ic->ic_state;
DPRINTF(("wi_newstate: %s -> %s\n", stname[ostate], stname[nstate]));
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
switch (nstate) {
case IEEE80211_S_INIT:
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (ic->ic_opmode != IEEE80211_M_MONITOR)
callout_stop(&sc->sc_rssadapt_ch);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ic->ic_flags &= ~IEEE80211_F_SIBSS;
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
sc->sc_flags &= ~WI_FLAGS_OUTRANGE;
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
return (*sc->sc_newstate)(ic, nstate, arg);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case IEEE80211_S_RUN:
Handle of AP_IN_RANGE/AP_OUT_RANGE, to fix a problem once the station is out of ragen, it never comes up again. For HostAP, prohibit sending DATA frame to an unassociated but authenticated station.
2002-10-04 08:23:20 +04:00
sc->sc_flags &= ~WI_FLAGS_OUTRANGE;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
buflen = IEEE80211_ADDR_LEN;
In wi_newstate(), read the BSSID into a temporary buffer that has guaranteed alignment. Fixes a really obscure failure mode on Alpha, and panics on other machines.
2003-10-25 03:58:22 +04:00
IEEE80211_ADDR_COPY(old_bssid.wi_mac_addr, ni->ni_bssid);
wi_read_rid(sc, WI_RID_CURRENT_BSSID, &bssid, &buflen);
IEEE80211_ADDR_COPY(ni->ni_bssid, &bssid);
IEEE80211_ADDR_COPY(ni->ni_macaddr, &bssid);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
buflen = sizeof(val);
wi_read_rid(sc, WI_RID_CURRENT_CHAN, &val, &buflen);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
if (!isset(ic->ic_chan_avail, le16toh(val)))
panic("%s: invalid channel %d\n", sc->sc_dev.dv_xname,
le16toh(val));
ni->ni_chan = &ic->ic_channels[le16toh(val)];
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
In wi_newstate(), read the BSSID into a temporary buffer that has guaranteed alignment. Fixes a really obscure failure mode on Alpha, and panics on other machines.
2003-10-25 03:58:22 +04:00
if (IEEE80211_ADDR_EQ(old_bssid.wi_mac_addr, ni->ni_bssid))
A Prism firmware participating in an IBSS network will not send a LINK_STAT notification for every change of BSSID, so the firmware's BSSID and the driver's BSSID will get out of sync. This has two bad affects. First, because the 802.11 framework filters received packets based on BSSID, many packets can be dropped before the driver adopts the firmware's BSSID. Second, ifconfig's tells a misleading BSSID to the operator. This problem is most apparent in networks where every station does not hear every other. I reproduce these conditions in an office by removing/replacing the antennas on my 802.11 cards. As a solution, in IBSS mode, the driver will ask the firmware for the BSSID every five seconds. Also, whenever the driver receives a frame carrying a different BSSID than the driver's BSSID, then the driver asks the firmware for the BSSID before handing the frame to ieee80211_input.
2003-01-01 05:06:47 +03:00
sc->sc_false_syns++;
else
sc->sc_false_syns = 0;
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
if (ic->ic_opmode == IEEE80211_M_HOSTAP) {
ni->ni_esslen = ic->ic_des_esslen;
memcpy(ni->ni_essid, ic->ic_des_essid, ni->ni_esslen);
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ni->ni_rates = ic->ic_sup_rates[
ieee80211_chan2mode(ic, ni->ni_chan)];
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
ni->ni_intval = ic->ic_lintval;
ni->ni_capinfo = IEEE80211_CAPINFO_ESS;
if (ic->ic_flags & IEEE80211_F_WEPON)
ni->ni_capinfo |= IEEE80211_CAPINFO_PRIVACY;
} else {
buflen = sizeof(ssid);
wi_read_rid(sc, WI_RID_CURRENT_SSID, &ssid, &buflen);
ni->ni_esslen = le16toh(ssid.wi_len);
if (ni->ni_esslen > IEEE80211_NWID_LEN)
ni->ni_esslen = IEEE80211_NWID_LEN; /*XXX*/
memcpy(ni->ni_essid, ssid.wi_ssid, ni->ni_esslen);
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
ni->ni_rates = ic->ic_sup_rates[
ieee80211_chan2mode(ic, ni->ni_chan)]; /*XXX*/
Fix NIC-memory leak for symbol firmware at reinitialization. After 4-5 times changing of 802.11 parameters, such as nwid, the error "tx buffer allocation failed" occured unless ifconfig down up. Pass signal streangth and timestamp to ieee80211_input, though it is not useful for wi driver for now.
2002-10-02 21:11:34 +04:00
}
Enable RSSI-based rate-adaptation for wi(4). This enables wi(4)-based APs to run at "full speed" where before they ran at just 1 or 2Mb/s. The AP will adapt the data rate for each client based on packet losses and the received signal strength. I have also enabled rate adaptation for STA mode and for IBSS mode, but the hardware gives us less control over the data rate in those modes.
2003-12-07 08:44:49 +03:00
if (ic->ic_opmode != IEEE80211_M_MONITOR)
callout_reset(&sc->sc_rssadapt_ch, hz / 10,
wi_rssadapt_updatestats, sc);
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case IEEE80211_S_SCAN:
case IEEE80211_S_AUTH:
case IEEE80211_S_ASSOC:
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
break;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
ieee80211_newstate() wants for ic->ic_state to equal the previous state, so do not update ic->ic_state in wi_newstate() before calling ieee80211_newstate(). Instead, update ic->ic_state when we do not let ieee80211_newstate() run.
2003-10-17 07:58:23 +04:00
ic->ic_state = nstate;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* skip standard ieee80211 handling */
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
return 0;
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Host AP power saving support. The Host AP notices that the power saving bit is set in incoming frames from a station, and buffers the outgoing frames for the station until they are polled for. This requires support in the driver to set a bit in the TIM bitmap sent during 802.11 beacons. So far, support for power saving in Host AP mode is only available for the PRISM2 chipset.
2003-04-08 08:31:23 +04:00
static int
wi_set_tim(struct ieee80211com *ic, int aid, int which)
{
struct wi_softc *sc = ic->ic_softc;
aid &= ~0xc000;
if (which)
aid |= 0x8000;
return wi_write_val(sc, WI_RID_SET_TIM, aid);
}
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
static int
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
wi_scan_ap(struct wi_softc *sc, u_int16_t chanmask, u_int16_t txrate)
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
{
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int error = 0;
u_int16_t val[2];
if (!sc->sc_enabled)
return ENXIO;
switch (sc->sc_firmware_type) {
case WI_LUCENT:
(void)wi_cmd(sc, WI_CMD_INQUIRE, WI_INFO_SCAN_RESULTS, 0, 0);
break;
case WI_INTERSIL:
From Steve Woodford (scw@): fix wi(4) and wiconfig(8) on big-endian machines.
2003-11-16 12:41:01 +03:00
val[0] = htole16(chanmask); /* channel */
val[1] = htole16(txrate); /* tx rate */
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
error = wi_write_rid(sc, WI_RID_SCAN_REQ, val, sizeof(val));
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
break;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
case WI_SYMBOL:
/*
* XXX only supported on 3.x ?
*/
val[0] = BSCAN_BCAST | BSCAN_ONETIME;
error = wi_write_rid(sc, WI_RID_BCAST_SCAN_REQ,
val, sizeof(val[0]));
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
break;
}
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
if (error == 0) {
sc->sc_scan_timer = WI_SCAN_WAIT;
sc->sc_ic.ic_if.if_timer = 1;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
DPRINTF(("wi_scan_ap: start scanning, "
"chanmask 0x%x txrate 0x%x\n", chanmask, txrate));
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
return error;
}
static void
wi_scan_result(struct wi_softc *sc, int fid, int cnt)
{
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
#define N(a) (sizeof (a) / sizeof (a[0]))
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
int i, naps, off, szbuf;
struct wi_scan_header ws_hdr; /* Prism2 header */
struct wi_scan_data_p2 ws_dat; /* Prism2 scantable*/
struct wi_apinfo *ap;
off = sizeof(u_int16_t) * 2;
memset(&ws_hdr, 0, sizeof(ws_hdr));
switch (sc->sc_firmware_type) {
case WI_INTERSIL:
wi_read_bap(sc, fid, off, &ws_hdr, sizeof(ws_hdr));
off += sizeof(ws_hdr);
szbuf = sizeof(struct wi_scan_data_p2);
break;
case WI_SYMBOL:
szbuf = sizeof(struct wi_scan_data_p2) + 6;
break;
case WI_LUCENT:
szbuf = sizeof(struct wi_scan_data);
break;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
default:
printf("%s: wi_scan_result: unknown firmware type %u\n",
sc->sc_dev.dv_xname, sc->sc_firmware_type);
naps = 0;
goto done;
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
}
naps = (cnt * 2 + 2 - off) / szbuf;
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
if (naps > N(sc->sc_aps))
naps = N(sc->sc_aps);
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
sc->sc_naps = naps;
/* Read Data */
ap = sc->sc_aps;
memset(&ws_dat, 0, sizeof(ws_dat));
for (i = 0; i < naps; i++, ap++) {
wi_read_bap(sc, fid, off, &ws_dat,
(sizeof(ws_dat) < szbuf ? sizeof(ws_dat) : szbuf));
DPRINTF2(("wi_scan_result: #%d: off %d bssid %s\n", i, off,
ether_sprintf(ws_dat.wi_bssid)));
off += szbuf;
ap->scanreason = le16toh(ws_hdr.wi_reason);
memcpy(ap->bssid, ws_dat.wi_bssid, sizeof(ap->bssid));
ap->channel = le16toh(ws_dat.wi_chid);
ap->signal = le16toh(ws_dat.wi_signal);
ap->noise = le16toh(ws_dat.wi_noise);
ap->quality = ap->signal - ap->noise;
ap->capinfo = le16toh(ws_dat.wi_capinfo);
ap->interval = le16toh(ws_dat.wi_interval);
ap->rate = le16toh(ws_dat.wi_rate);
ap->namelen = le16toh(ws_dat.wi_namelen);
if (ap->namelen > sizeof(ap->name))
ap->namelen = sizeof(ap->name);
memcpy(ap->name, ws_dat.wi_name, ap->namelen);
}
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
done:
Jumbo commit for wi driver. - Eliminate wi_hostap.c since most of the code are duplicated with net/if_ieee80211subr.c - Station for Infrastructure network and IBSS also use service functions as much as possible to be consistent with other wireless drivers. Now WEP works for station/ibss/hostap.
2002-09-30 10:50:35 +04:00
/* Done scanning */
sc->sc_scan_timer = 0;
DPRINTF(("wi_scan_result: scan complete: ap %d\n", naps));
Begin synchronization with FreeBSD: 1 Add channel mask, transmit rate-mask arguments to wi_scan_ap. 2 Adopt the macros WI_LOCK/WI_UNLOCK for synchronization. FreeBSD uses a different synchronization mechanism. 3 In wi_cmd, adopt constants WI_DELAY/WI_TIMEOUT for timing. 4 Pull debug messages from into wi_read_nicid from FreeBSD. 5 Bug fix: if IFF_ALLMULTI, don't filter any multicasts. 6 Count and report TX exceptions, but don't generate any additional interrupts.
2003-05-13 12:35:58 +04:00
#undef N
Jumbo patch, from David Young <dyoung@ojctech.com>, with small tweaks by me: * Speed up reading/writing buffers from the hardware by avoiding slow forward seeks. In preparation to use the optimization, do not read overlapping bytes. This is currently disabled, but can be enabled with OPTIMIZE_RW_DATA. * Hand 802.11 and Prism-specific frames to BPF. User can watch these frames by specifying an alternate DLT to e.g. tcpdump(8). * Add support for SIOC[SG]80211BSSID and SIOC[SG]80211CHANNEL. * Issue join requests and track join/create state through link-status notifications. * Split wi_rxeof into separate routines for receiving Ethernet II, 802.11 data, and 802.11 management frames. * Bug fix: Account for aligning m_data to a word boundary in the Rx buffer size check. * Bug fix: Check for LLC/SNAP even if the firmware tells us the frame is Ethernet II, as the firmware sometimes gets this wrong. * Process as many events as possible when we get an interrupt, using a simple heuristic to avoid reprocessing an event (which can have bad side-effects). Clamp the time spent in the interrupt handler to 4ms. * Redo the timeout loops to be consistent and less prone to error. * Add delays to timeout loops which were missing them, so that a fast CPU won't win the race. * Borrow some timeout loop values from the linux-wlan-ng driver, which seems to reflect a high level of clue (due to direct support from Intersil). * Get rid of silly wi_read_data(..., len + 2) idiom; simply round up in wi_read_data() and wi_write_data(). Also, protect against a length of 0. * Name some frequently-used constants. Correct spelling. Other style nits. * Bug fix: On Prism, set Create IBSS register to 0 *always*. The meaning of Create IBSS == 1 is join an IBSS or *ESS*, and we do not want to join an ESS, because that would put us in an inconsistent state. 0 is the right value for Prism. * Bug fix: Clean up state at the top of wi_init(), in the event that we don't reach the bottom. * Simplify wi_start() by always providing an RFC1042-encoded 802.11 frame to the firmware. * Larval powersave support for HostAP mode, enabled by WI_HOSTAP_POWERSAVE. * Bug fix: Call wi_stop() from wi_shutdown(). * Bug fix: sync media options with HostAP mode in wi_sync_media(). * In wi_media_status(), inquire firmware for current media state if media == auto. From FreeBSD. * Clean up the way buffer lengths are computed by using pointer arithmetic rather than magic constants. * Swap the order of comparisons in addr_cmp() for speed. * Bug fix: Send ReAssoc Response instead of Assoc Response to a ReAssoc Request. * Bug fix: Copy SSID using the correct size. * Give more meaningful names to offsets in a wi_frame. * Bug fix: Assign the right values to the named constants for Rx frame encoding. * Get rid of useless SNAP constants.
2002-09-23 18:31:27 +04:00
}
Debug code from FreeBSD: handy packet dumps when IFF_DEBUG|IFF_LINK2 are set.
2003-05-13 10:48:56 +04:00
static void
wi_dump_pkt(struct wi_frame *wh, struct ieee80211_node *ni, int rssi)
{
ieee80211_dump_pkt((u_int8_t *) &wh->wi_whdr, sizeof(wh->wi_whdr),
Switch wi(4) to the new 802.11 layer.
2003-10-13 12:07:21 +04:00
ni ? ni->ni_rates.rs_rates[ni->ni_txrate] & IEEE80211_RATE_VAL
: -1,
rssi);
Debug code from FreeBSD: handy packet dumps when IFF_DEBUG|IFF_LINK2 are set.
2003-05-13 10:48:56 +04:00
printf(" status 0x%x rx_tstamp1 %u rx_tstamp0 0x%u rx_silence %u\n",
le16toh(wh->wi_status), le16toh(wh->wi_rx_tstamp1),
le16toh(wh->wi_rx_tstamp0), wh->wi_rx_silence);
printf(" rx_signal %u rx_rate %u rx_flow %u\n",
wh->wi_rx_signal, wh->wi_rx_rate, wh->wi_rx_flow);
printf(" tx_rtry %u tx_rate %u tx_ctl 0x%x dat_len %u\n",
wh->wi_tx_rtry, wh->wi_tx_rate,
le16toh(wh->wi_tx_ctl), le16toh(wh->wi_dat_len));
printf(" ehdr dst %s src %s type 0x%x\n",
ether_sprintf(wh->wi_ehdr.ether_dhost),
ether_sprintf(wh->wi_ehdr.ether_shost),
wh->wi_ehdr.ether_type);
}
Reference in New Issue Copy Permalink