Changes in 1.2.1

* Fix driver not make associate request when required
* Fix iwconfig essid any doesn't associate problem
* Fix ipw_associate_network() debug log bug
* Fix ifconfig reports Link-encap as Ether (thanks to Jason)
* Fix soft lockup during driver load time
* Add Makefile warning if a different SHELL other than bash is used
* Use the correct variable to print the direct scan essid
* Fix a kernel Oops bug #1248
* Fix hardware/software RF kill switch problem during module loading time

Changes in 1.2.0

* Workqueue rework for kernel 2.6.20 and above
* WE-21 ESSID fix
* Event log output enhancement

Changes in 1.1.4

* Fixed reversed version logic in snapshot/compat_ieee80211
* Removed unused usage of ieee80211_action to help building in older
  kernels
* Modified build so if KSRC is provided, /lib/modules/$(uname) is
  not searched for ieee80211 files.
* Fixed compile problem on Dapper (thanks to Kyle McMartin)

Changes in 1.1.3

* Fixed problem with newer kernels and config.h vs. autoconf.h
* Fixed problem with IEEE80211 API compatibility build issues.
* Fixed #1096 problem with soft lockup while loading if RF kill active.
* Fixed compile warning ignoring pci_enable_device's return code.
* Added retransmission of association and authentication requests if no 
  response received.
* Changed the make target install_for_testing to install.

Changes in 1.1.2

* Fixed #972 problem with wpa_supplicant not selecting hidden APs
* Fixed #1113 problem with compiling against ieee80211-1.1.14
* Fixed #1146 problem with last character of SSID being dropped.
* Fixed #1089 circular locking dependency during register_netdev
* Fixed #1090 problem with setting rate while not associated.
* Fixed #1098 possible memory corruption on 64-bit systems

Changes in 1.1.1

NOTE: Since 1.1.0-pre2 was elevated to 1.1.0, the unstable series 
      version was updated to 1.1.x.

* Updated README.ipw3945, ISSUES, and INSTALL to be in sync with 1.1.0.
* Modified to not filter out legacy (non-802.11h) access points on 
  channels marked as being in the radar detect spectrum.  You must 
  upgrade to the ipw3945d regulatory daemon version 1.7.22 or newer in 
  order to take advantage of this functionality.

Changes in 1.0.12

* Added support for new ieee80211 API version #defines (forward 
  compatibility)
* Added snapshot/check_kernel for 'make patch_kernel' support

Changes in 1.0.11

* Fixed problem with compiling against ieee80211 API v1 (missing a ;)
* Added more output during start of build to help debug problems 
  encountered by some users when running make.
* Fixed #1062 problem with rx_packets incrementing without traffic.
* Fixed #1078 problem with privacy bit not broadcasting correctly in 
  IBSS beacons.
* Sync with 1.1.0-pre2 cleanup changes (fixed sim_rx build and added 
  some additional ieee80211 API version checking during build)

Changes in 1.0.10

* Fix #1067 problem with compiling against ieee80211-1.1.14.
* Modified Makefile to detect ieee80211-1.1.14 and define 
  IEEE80211_API_VERSION to be 2.
* Modified Makefile to detect if your kernel provides ieee80211 symbol 
  definitions in Module.symvers but the driver build will be using a 
  different ieee80211 subsystem.
* Fixed CONFIG_IPW_QOS => CONFIG_IPW3945_QOS defines in Makefile

Changes in 1.0.9

* Fixed problem with mandatory basic rates not being added correctly into
  association requests if they have been masked out by a mode selection.

Changes in 1.0.8 

* Fixed #1059 problem with iwlist rate not working.

Changes in 1.0.7

* Fixed problem with 802.11a channels not associating.
* Fixed behavior with short_slot_time and short_preamble capabilities

Changes in 1.0.6

* Fixed problem with driver requesting masked rates during association 
* Changed Makefile to attempt to find the ieee80211 subsystem and to 
  prompt the user if it is not found.  If it is found outside of the
  default kernel build location then the MODVERDIR parameter is passed
  to the modules build target.

Changes in 1.0.5

* Added check_ieee80211_version and find_ieee80211 scripts required for 
  build.

Changes in 1.0.4

* Fixed #1035 problem with not being able to associate with active key 
  not being set to [1].
* Fixed #1043 problem with periodic system hang while switching between 
  BSS and IBSS modes.
* Fixed #1024 problem with associating to open AP after WPA APs

Changes in 1.0.3

* Fix #1021 problem with using uninitialized crypt pointer during
  shared-key authentication.

Changes in 1.0.2

* Fix #1025 problem with not being able to trun off RTS via iwconfig.

Changes in 1.0.1

* Fix #1018 problem with BUG_ON if duplicate packet recieved w/out retry bit.
* Fixed a type-o in debug messages (s/3946/3945/)
* Set enabled but non-supported features to default to n in Kconfig.
* Added util/Makefile to build sim_rx
* Modified Makefile so that if the 'make clean' in the util directory
* Fix problem with not setting privacy bits correctly.
* Fix problem with using uninitialized bssid during association.
* Ran scripts/Lindent on ipw3945.{c,h}
* Fix #934 to correctly clear out station table when switching modes.
* Fix #1013 to properly set capability bits.
* This patch adjusts the rate scaling code to flush the entries at a more
* Set RTS handling to match 802.11 spec
* Stamped version string with 'k' to indicate it is not an actual
* Ran /usr/src/linux/scripts/Lindent on ipw3945.{c,h}
* Added simulation Rx sample application to util directory.

For additional detail on above changes, see:
	http://bughost.org/repos/ipw3945.git/

Changes in 1.0.0

NOTE:  v0.0.69 completed a validation pass and was elevated to 1.0.0.  

Changes in 0.0.74

* Fixed #994 Modified Makefile to only add the CONFIG_* definitions to 
  CFLAGS if the corresponding Makefile variable is set to enabled, 
  otherwise the #ifdef checks in the C code will always evaluate to 
  true. Also wraps the debug module parameter in #ifdefs (thanks to 
  Henrik Brix Andersen)
* Fixed #987 (for the 3945) to not use MODVERDIR when running make 
  (thanks to Henrik Brix Andersen)
* Fixed #980 problem with iwconfig ethX power not being able to set 
  power to ON.
* Fixed problem with iwlist scan returning that a scan was in progress 
  while trying to associate.
* Fixed filters script to look in the correct location for the filters 
  file
* Changed default retry per rate count from 2 to 1 to improve 
  throughput.
* Added sysfs entry to manually select new retry pre rate counts 

Changes in 0.0.73

* Fixed problem with RSSI having IPW_RSSI_OFFSET being subtracted twice 
  (resulting in *very low* signal levels)

Changes in 0.0.72

* Fixed #972 problem with not being able to associate to hidden SSIDs 
  using wpa_supplicant.
* Fixed #976 problem with RTS threshold value being clamped at 2304 vs. 
  2344.
* Fixed #933 problem with association to hidden networks timing out 
  (increased the timeout from 500ms to 750ms)
* Merged patch to set 'fixed' flag for the WE get rate handler (thanks 
  to Jean Tourrilhes)
* Added rtap_filter sysfs entry for configuring which packets are 
  sent to user space via the rtap interface.
* Modified the tx and rx byte count statistics to count only 802.11
  data payloads vs. counting miscellaneous 802.11 overhead.
* Changed single usage of kfree_skb to dev_kfree_skb_any to fix a 
  compile problem seen by some users.
* Fixed problem with timed out host commands reporting the wrong 
  command # to the debug log.
* Added code to wait for an in-progress scan to abort before 
  switching modes.
* Modified scan request to block until the request is acknowledged 
  before returning to ensure that the scanning state machine is in sync 
  with the hardware state machine.

Changes in 0.0.71

* Fix #964 problem with disassociation on hidden networks
* Fix #956 problem loading driver if RF Kill switch enabled in 0.0.70
* Fixed problem with switching from BG to A mode via iwpriv set_mode 
  while in an ad-hoc network.
* Change association timeout value from HZ/2 to 750ms (see #933)
* Added rtap interface for RF promiscuous mode while associated.  
  Enabled via CONFIG_IPW3945_RADIOTAP.
* This adds the ability to simulate reception of a frame to the driver 
  to test code paths. Enabled via CONFIG_IPW3945_SIM_RX.

Changes in 0.0.70

* Modified rate scaling to not send an unnecessary command to the
  microcode every time the rate changes.
* Modified the Tx measurement code to only gather statistics once every
  3 seconds vs. every time a packet exceeds a given size (relative to 
  the rate used)
* Fixed #932 - Modified Tx measurement code to disable BT priority for those 
  Tx requests where measurements are being requested (thanks to Eric Shoemaker
  for helping investigate and test patches)

Changes in 0.0.69

* Fixed #295 Removed BUG_ON(!spin_is_locked(&priv->daemon_lock)) code 
  as spin_is_locked is always 0 on UP and non-debug kernels.

Changes in 0.0.68

* Updated README.ipw3945:
	- Power Managment section reflect actual operation of the 
	  driver.
	- Updated IEEE 802.11h information
	- Updated 'channels' information in sysfs section
* Fixed copyright date in driver #define to 2006 (from 2004)
* Fixed driver name.
* Updated ucode and regulatory daemon version strings in documentation.

Changes in 0.0.67

* First version posted to SourceForge
* Changed QoS module parameters to all have the qos_ prefix.
* Updated INSTALL and REAMDE.ipw3945 documents.
* Moved CONFIG_IPW_MONITOR -> CONFIG_IPW3945_MONITOR
* Moved reset priv out of CONFIG_IPW3945_MONITOR mode ifdef's
* Defined all symbols as static

Changes in 0.66

* Fix #279 problem with driver not supporting BG SKU by adding support for 
  BG SKUs with the following PCI ID (device:subsystem)

    0x4222:0x1005
    0x4222:0x1034
    0x4227:0x1014
    0x4222:0x1044

  If a card is detected as ABG (its PCI ID is not one of the above) and 
  the EEPROM does not enable any A band channels, it will be treated as 
  a BG SKU and an informational message will be displayed to the kernel 
  log.
* Fix #267 problem with comapre_ether_addr being redefined by ipw3945 
  after ieee80211 has already done so on kernels older than 2.6.15

Changes in 0.65

* Fix #260 problem with frames setting PM bit even after power 
  saving mode has been disabled.
* Updated documentation to fix #172, #180, #189, #124, and #249
* Fix #264 problem with first load after a restart 
  potentially causing the intial load to be delayed due to 
  the NIC not being active when it an attempt is made to 
  take it down:
	- Changed !(priv->status & STATUS_INIT) -> ipw_is_alive
	- Changed ipw_is_init -> ipw_is_alive 
	- Changed STATUS_INIT -> STATUS_ALIVE
	- Added STATUS_INIT to indicate if ipw_nic_init was called
	- Added check in ipw_down for whether STATUS_INIT is set
* Fix #265 problem with suspend while associated 
  periodically causing a kernel panic.
* Fix #266 problem with resume process taking a long time 
  and causing keyboard to be unresponsive.

Changes in 0.64

* Fix #257 problem with daemon/driver protocol problems during error
  recovery as a result of adapter_restart.
* Fix #259 problem with 2.4Ghz networks supporting only OFDM rates being 
  viewed as 5.2Ghz.
* Fix #255 problem with 'error -512' being displayed during suspend.  
  Several fixes went in related to suspend/resume work.
* Fixed #259 problem with OFDM only network appearing as 5.2Ghz
* Fixed #250 wireless tools race condition with driver during 
  initialization.
* Fixed problem with LED not turning off if RF KILL activated.

Changes in 0.63

* Fixed #246 problem with iwconfig not reflecting current txpower 
  setting.
* Fixed #247 problem with iwconfig not reflecting current maximum rate
  setting.
* Fixed #251 problem with driver sometimes not recovering from a HW 
  RF Kill switch toggle.
* Fixed #248 problem with passing static channel configuration as module 
  parameter resulting in a BUG_ON.
* Fixed #252 problem with driver not being able to fully 
  re-initialize itself when loaded.
* Removed printk_buf from non-debug builds

Changes in 0.62

* Fixed problem with maximum txpower limit being set to 0, which keeps 
  iwconfig txpower from functioning.
* Fixed #214 by providing ability to disable roaming via sysfs entry 'roam'.
  Setting to 0 disables roaming.
* Modified init sequence to defer sending the INIT to the daemon until 
  the hardware is reporting valid temperature readings.

Changes in 0.61

* Fixed #239 problem with HW rf kill not working
* Fixed #127 problem with associating with 802.11a APs that have the 
  spectrum management bit set but were not providing Power Constraint 
  and TPC report information elements.
* Fixed #114 problem with killing ipw3945d while associated causing
  firmware errors.
* Fixed #132 stack usage problem.  Any function using > 300 
  bytes of stack space was modified to use less stack space.  
  The largest stack usage is now < 300 bytes.
* Fixed #128 and #157 problems with noise stats being dropped 
  to 0 if CCK probe frame is receivd on an OFDM network.
* Changed priv->sem to priv->mutex based on kernel inclusion 
  requirements.

Changes in 0.60

* Fix #243 - Added code to ensure all timers, delayed work, and work 
  queues have completed execution before unloading the driver module.
* Fixed dead lock if Tx ISR came in while the daemon_lock was held and 
  the command completion callback kicked in (recursive dead lock)

Changes in 0.59

* Added code to ensure all timers, delayed work, and work queues have 
  completed execution before unloading the driver module.
* Added code to force all driver entry points to immediately return if 
  a driver shutdown is in progress

Changes in 0.58

* Fixed dead lock if Tx ISR came in while the daemon_lock was held and the 
  command completion callback kicked in (recursive dead lock)
* Added code to abort a ipw_daemon_send_cmd() if an unload is initiated

Changes in 0.57

* Fixed possible problem with DAEMON_READ_TIMEOUT length not being
  initialized if compiled with certain versions of gcc; symptom is
  the message 'read mistmatch: RANDOM NUMBER (expected) !=0' in the
  ipw3945d daemon debug log (originally listed as fixed in 0.55, but 
  not actually merged into code).
* Possible fix for spin_lock dead lock during heavy Tx activity
* Fixed problem with sending daemon Tx measured status even on on Tx 
  failures

Changes in 0.56

* Fixed #242 problem with firmware restarts in 802.11a mode
* Added Tx rate information to the Tx daemon status sync.

Changes in 0.55

* Fixed #195 problem with iwconfig txpower not setting or reporting the
  configured Tx power correctly.  Documented the expected behavior in 
  README.ipw3945.
* Fixed #236 problem with inability to reassociate with WPA APs without
  reloading the driver.
* Fixed #190 problem with bitrate not being reported correctly as a 
  function of range
* Fixed #210 problem with PSP mode not working correctly in BSS mode 
  (not quite what the summary description of the bug was, but that's 
  where the bug fix led)
* Fixed #241 problem with 'Error sending DAEMON to daemon' message if
  you unload the driver while associated.

Changes in 0.54

* Fixed problem with 2.4Ghz APs not associating
* Fixed problem with scans entering deadlock if a second scan request 
  is scheduled before a prior one has initialized the hardware.

Changes in 0.53

* Fixed problems with rate scaling (should improve all performance 
  related bugs)
* Fixed problem with SW rf_kill not being able to turn the adapter 
  back on
* Added plumbing for doing Tx measurement reporting from the driver to 
  the daemon.

Changes in 0.52

* Fixed load/unload scripts so they correctly pass parameters (unload 
  was resetting the driver and daemon arguments to an empty list at the 
  start of the load script)
* Fixed #102 problem with not being able to associate in Managed mode 
  after using Ad-Hoc mode
* Removed bit-field from being displayed via channels sysfs entry.
* Added spin_lock mutex around daemon user space and cross thread 
  communication constructs
* Fixed problem where if a scan abort is received after the microcode
  has completed scanning but before we have been notified, scanning 
  would no longer function.

Changes in 0.51

* First external release
* Cleaned up QoS code so it is all wrapped behind CONFIG_IPW3945_QOS
* Changed several #define clusters into enum { } sections, replacing
  numeric constants with bit-shift constants.
* Fixed problem with 'make patch_kernel' not copying the driver source 
  files into the kernel tree.

Changes in 0.50

* Added 'patch_kernel' support to the Makefile so you can build and run 
  in-tree.
* Fixed problems with load and unload script not being able to find 
  ipw3945d in /sbin or via the -ipw3945d= load script parameter.
* Updated documentation on using -ipw394d= parameter

Changes in 0.49

* IRC only release to test the load / unload scripts.

Changes in 0.48

* Added synchronous daemon command capabilities
* Changed API for ipw_daemon_send_cmd() -- caller must allocate storage for 
  the daemon command.  Once called, ipw_daemon_cmd assumes control of the
  storage.  The caller must not free or reference the data once called, 
  regardless of ipw_daemon_send_cmd return code.
* Added daemonqueue as a 3rd execution thread.  This thread is scheduled
  to processes command requests from the daemon into the driver.
* Added DAEMON_CMD_DONE for the daemon to send results to the driver when
  a synchronous driver initiated command has completed.
* Moved post-calibration work to be scheduled to execute in priv->workqueue 
  vs. in the user thread used to call store_cmd().

Changes in 0.47

* Changed output of channels sysfs entry again.
* Fixed A and B rate scaling code
* Modified load and unload scripts to not specify the patch for ipw3945d 
  -- they should be installed somewhere in your path.  If you do not 
  have ipw3945d installed in your path, you can pass the -ipw3945d= 
  parameter to the load script to point to the directory where ipw3945d 
  can be located, for example:
	% ./load -ipw3945d=~/bin/ debug=0x43fff -- --log-file=ipw3945d.log
* Added daemon / driver version checking.

Changes in 0.46

* Changed sysfs channel output to be more compact and detail more info.
* Major reduction in code #defines and enums, #if 0 / #endif, unused commands,
  etc.  

Changes in 0.45

* Fixed problem with not being able to associated with shared-key 
  authentication
* Fixed problem with probe request not being 802.11 spec compliant, 
  resulting in direct scans not working with some APs.

Changes in 0.44

* Fixed problem with LED rate not changing while scanning.
* Fixed problem with LED state not turning off if rf_kill asserted

Changes in 0.43

* Fixed problem with driver not compiling if WIRELESS_EXT < 18
* Fixed problem with probe requests sending incorrect packet length 
  (introduced with 0.42)
* Added /sys/bus/pci/drivers/ipw3945/*/channels to provide a user 
  indication of which channels their SKU supports.

Changes in 0.42

* Fixed problem with background scans (scanning while associated) 
  periodically rendering the adapter inoperable (requiring reload)
* Fixed direct/indirect scan sequencing so that only a single scan pass
  is required per scan
* Wrapped all QOS code in CONFIG_IPW3945_QOS (defaulting to undefined --
  compiled out)

Changes in 0.41

* Fixed problem with not reassociating after resume from susp-to-disk
* Fixed #225 problem with NETDEV_TX_BUSY message in kernel log

Changes in 0.40

* Fixed #201 problem with not being able to associate after switching 
  from IBSS mode
* Fixed #204 problem with setting channel after going into Ad-Hoc mode
* Fixed #213 problem with txpower being reset upon association to 
  802.11h AP

Changes in 0.39

* Added TKIP countermeasure fixes based on code in more recent versions 
  of ipw2200

Changes in 0.38

* Fixed #200 problem with enabling LEDs causing a kernel oops on 
  association.  See README.ipw3945 for the intended behavior of the 
  LEDs.
* Fixed #203 problem with LEDs defaulting to being off.

Changes in 0.37

* Fixed #58 problem with datatransfer speed being too low in IBSS
* Fixed #106 and #146 problem with rate scaling dropping to '0' in IBSS
* Fixed #178 problem with fw error on line 74 durring IBSS association
* Fixed #199 problem with rate scaling not working if a fixed rate is 
  specified.
* Fixed #112 problem with not being able to associate after changing 
  mode from ad-hoc using a different channel.
* Fixed #162 problem with changing rates while associated causing 
  disassociation.

Changes in 0.36

* Make antenna selector a member of priv structure, to support multiple 
  adapters.
* Properly interpret antenna values sent via sysfs.
* Add support for the txpow command from iwconfig to correctly set Tx 
  limits.
* Added code to ignore receipt of multiple association requests if one 
  has already been processed for the current network.
* Fixed problem with probe requests not being sent after associating
* Changed scan timings so we can perform shorter scans if active scanning is
  workings (30ms active dwell, 20ms active quiet with a 1 packet threshold)
* Fixed minor bug affecting ipw_report_frame() debug message

Changes in 0.35

* Fixed BUG() condition in show_cmd that could cause a system lockup.
* Added sysfs entry for controlling antenna diversity (see 
  README.ipw3945) and removed code forcing antenna toggling during 
  scans. 
* Fixed problem where hidden networks were disassociating due to 
  missing beacons.

Changes in 0.34

* Fixed time reported in "Network exluded because of age" message.
* Enhances debug messages for Rx frames, improves "hackability" of Rx
  debug messages.
* Discard frames with *FIFO* overrun errors as well as CRC errors.
* Added code to allow driver to request a true passive scan (without
  auto-switch from passive to active)

Changes in 0.33

* Fixed printf() warning for specifying %d vs. %zd for size_t type.
* Fixed rate scaling problems
* Fixed #185 problem with sleep being called from invalid context 
  (ieee80211 subsystem)
* Fixed #176 problem with quiet request not being enforced
* Fixed #138 problem with associating to 5.2Ghz band after 2.4Ghz band 
  causing association delays.

Changes in 0.32

* Removed a stray antenna diversity investigation test that snuck into 0.31

Changes in 0.31

* Fixed problem with OOPS on measurement request

Changes in 0.30

* Fixed another startup race condition where a scan request issued too 
  soon results in no scans occurring.

Changes in 0.28

* Fixed #157 problem with noise stats not being reported via iwconfig.  
  NOTE:  noise statistics are only available when receiving OFDM frames.
* Fixed #166 problem with 802.11h not functioning (inability to associate 
  with 802.11h enabled access points)

Changes in 0.26

* Fixed #155 problem with commands sent to driver too quickly after 
  load not resulting in association (no scan was being automatically 
  kicked off after the load cycle was complete)
* Fixed problem with 'config' script not functioning

Changes in 0.24

* Fixed ucode version reporting through ethtool
* Fixed a potential hang on rmmod due to unkilled timers
* Added a lot of debug dumps for scanning (trying to root cause 802.11a 
  association  issue)

Changes in 0.23

* Fixed OOPS on rmmod due to referencing list element after freeing it.
* Fixed potential command ring buffer overflow (may have been causing 142)
* Fixed potential problem with daemon trying to send commands before fully 
  initialized
* Fixed #137 problem with not being able to exit shells after loading 
  daemon.
* Fixed #147 problem with changing AP to RF overlapping channels never 
  disassociating.

Changes in 0.22

* Fixed problem with load script not launching ipw3945d.
* Fixed problem with load time race condition.

Changes in 0.20

* Improved #12 to better report statistics when available from the 
  uCode (signal and noise stats are not always provided from the 
  uCode)
* Fix problem with switching APs to a new channel w/in RF overlap range 
  not resulting in client following the AP.
* Misc. documentation updates to the source code
* Fix problem with roaming scanning on all channels EXCEPT the current
  channel, which results in it never finding a network to roam to.
* Added additional debug traces for restart conditions at IPW_DEBUG_INFO
  debug level.
* Fixed problem with antenna diversity (it wasn't cycling between antennas)
* Modified the DEAUTH handler to disassociate to fix a problem with the 
  driver becoming inoperable w/out a reload if a DEAUTH frame is received
  while associated (please retest #56)

Changes in 0.19

* Fixed #83 frequent firmware errors after association

Changes in 0.18

* Fixed WE18 compile problem

Changes in 0.17

* Fixed #94 error message in syslog when ipw3945d killed.
* Fixed #88 problem with unload script exiting shell

Changes in 0.16

* Fixed #12 problem with statistics not being reported correctly
* Fixed #38/57 problem with RF kill not working -- be sure to run 
  ipw3945d after excersizing RF kill
* Fixed #56/92 problem with DEAUTH
* Fixed #73 problem with 'iwlist retry' not working
* Fixed #97 problem with association in ad-hoc not working

Changes in 0.15

* Fix #84 module unloading failing due to oops
* Fix #81 type-o in debug output for 'association'
* Fix #82 add text interpretation of first field of fw restart
* Changed several messages that were incorrectly using IPW_WARNING (vs. 
  just IPW_DEBUG_INFO or other maskable debug level)
* Modified behavior of load and unload to manage the ipw3945d daemon

Changes in 0.13

* Fix #72 problem with associating with APs not broadcasting 
  their ESSID

Changes in 0.12

* Fix #70 problem with using an unconfigured key causing a panic
* Fix #74 problem with iwpriv causing a panic
* Fix #61 problem with ad-hoc mode and encrypted networks

Changes in 0.11

* Fix #55,#59,#64 problem with lockups, hangs, and applications 
  disappearing due to a double free / null reference in the 
  authentication handler
* Fix #65 problem with iwconfig not showing disassociation
* Fix #70 problem with using an unconfigured key causing a panic
* Fix #73 problem with iwlist retry not reporting correct values

Changes in 0.10

* Fixed #60 README.ipw3945 contains wrong description

Changes in 0.9

* Fixed #33 problem with iwconfig retry not working.
* Fixed #43 problem with remove-old removing a USB driver.
* Fixed #50 problem with TKIP causing a kernel panic.
* Workaround for #49 problem with compiler not initializing data to 0.
* Fixed #52 problem with 'make install' -- installation of 
  non-production versions is not encouraged and is disabled by default.

Changes in 0.8

* Fixed #10 problem with index for DMA queue out of range (scp stall)
* Fixed #36 problem with different modes associating
* Fixed #26 association timeout when using restricted mode
* Fixed #38 RF Kill causes firmware restarts
* Fixed problem with ad-hoc networks merging 

Changes in 0.7

* Fix #24: Kicking off scan before rx_config called.


Changes in 0.6

* Fix #27: Applied fix related to IBSS and Roaming.


