spelling
feedback and ok tb@ jmc@ ok ratchov@

Remove wireless turbo mode support. It is a non-standard extension
which only worked with ath(4) devices from a decade ago.
Diff tested on ath(4) hardware by me to verify that 11a/b modes still work.
ok reyk deraadt chris sthen kettenis

add a size argument to free. will be used soon, but for now default to 0.
after discussions with beck deraadt kettenis.

replace custom macro AR5K_ELEMENTS() with nitems()

const void foo() makes no sense. No binary change.

ok niklas@ otto@

add a HAL function to force disabling the hardware crypto engine

introduce new IEEE80211_STA_ONLY kernel option that can be set to
remove IBSS and HostAP support from net80211 and 802.11 drivers.
it can be used to shrink RAMDISK kernels for instance (like what
was done for wi(4)).
it also has the benefit of highlighting what is specific to IBSS
and HostAP modes in the code.
the cost is that we now have two code paths to maintain.

backout last "fix" because it breaks operation on some devices. even
if the fix looks right, it seems to expose another problem that i
haven't found yet.

figured out by Denis Doroshenko at gmail dot com

fix the setup of Tx descriptors. the frame type and "no ack" bits were
initialized in ar5k_ar521*_setup_tx_desc() but cleared in
ar5k_ar521*_fill_tx_desc() by setting the segment length incorrectly.

From ath5k via Nick Kossifidis (mickflemm at gmail dot com)*

---snip---
The missing no_ack in broadcast frames caused them to be retried up to
the retry_limit(1+4=5 transmissions by default).
---snap---

* it was explicitly mentioned that it is ISC-licensed now

``it's'' -> ``its'' when the grammar gods require this change.

add some HAL functions (again) inspired by the changes from the
madwifi-old-openhal project. this was backed out the last time because
it was too close to the release and i didn't respect the ABI/API lock.

do not use bcopy to set the mac address registers. this diff should
fix the "hal" on big endian systems.

thanks to the madwifi-old-openhal people

back out, reyk is not paying attention

Add some HAL functions.

From madwifi-old-openhal via Nick Kossifidis (mickflemm at gmail dot com)

Bump copyright while I'm here.

Consistently spell FALLTHROUGH to appease lint.
ok kettenis@ cloder@ tom@ henning@

Update my e-mail address in the copyright statement, no binary changes.

KNF

fix a bug in Tx descriptor processing.

found by David Malone

don't attach unsupported radio chipsets

based on a diff by jsg@

ansi, no binary change.
ok reyk@

catched->caught

reduce stack usage

spotted out by jsg@

remove non-constant array initializers which is a GNU extension.
suggested by anil some time ago.

remove some misplaced htole32s

ath_hal_rx_monitor should be used for link state monitoring instead of
setting the promisc mode. rename it to ath_hal_rx_signal and remove
the incorrect promisc flag. this should fix some problems with ath
running in forwarding mode.

sync power management settings. this should fix some unrecoverable
sleep mode timeouts for ar5210, ar5211 and ar5212.

the ar5210 is silicon version 0, not 1.

fix silicon version number reporting. ie. ath0: mac 80.6 -> ath0: mac 5.6.

beautify the code by renaming HAL functions with capitalized words (i
always wanted to do that). this breaks HAL compatibility but porting
should be easy, have a look at athvar.h. no functional changes.

move reset of HAL descriptors to ath.c

bye bye unportable bitfields

pad keys with zeros to a supported key length. this is bad but seems
to be the default WEP behaviour for string-based keys and allows
commands like "ifconfig ath0 nwkey yuck" to work...

some fixes for wep and reset.

btw.: this is a commit using an ar5212 in 11b mode...

ok deraadt@, krw@

ignore bmiss interrupts for now.

fix beacon handling

a new year and a new, less restrictive license. because it has to be
free.

go ahead miod@

knf

cosmetic changes, fix regdomain code and beacon handling

fix an invalid condition which returns on an invalid beacon state.

fix regulation domain handling

sync with the latest work and add some stuff needed by the upcoming ar5211/ar5212
support. some further cleanups and changes will follow.

store number of supported gpio pins

some fixes for beeing compatible with gcc3 as noticed by some people.
thanks to Joerg Sonnenberger and Nathan Binkert.

use __packed instead of __attribute__ ((__packed__)) (ok miod@), check
possible NULL mac in the keycache function.

knf, use positive error values (suggested by miod@)

spacing, beautify dmesg output

removed some printfs

import of a free hal part for the ath driver as a replacement for the
binary-only hal module found in FreeBSD and NetBSD. OpenBSD's approach
is based on reverse engineering because it is _not_ possible to
include a non-free and binary-only piece of software in a 100% free
operating system. it still lacks some features found in the "official" hal
module but this will be done very soon with a help by a lot of
contributors - because it's free.

ok deraadt@

Remove wireless turbo mode support. It is a non-standard extension
which only worked with ath(4) devices from a decade ago.
Diff tested on ath(4) hardware by me to verify that 11a/b modes still work.
ok reyk deraadt chris sthen kettenis

add a size argument to free. will be used soon, but for now default to 0.
after discussions with beck deraadt kettenis.

replace custom macro AR5K_ELEMENTS() with nitems()

const void foo() makes no sense. No binary change.

ok niklas@ otto@

add a HAL function to force disabling the hardware crypto engine

introduce new IEEE80211_STA_ONLY kernel option that can be set to
remove IBSS and HostAP support from net80211 and 802.11 drivers.
it can be used to shrink RAMDISK kernels for instance (like what
was done for wi(4)).
it also has the benefit of highlighting what is specific to IBSS
and HostAP modes in the code.
the cost is that we now have two code paths to maintain.

backout last "fix" because it breaks operation on some devices. even
if the fix looks right, it seems to expose another problem that i
haven't found yet.

figured out by Denis Doroshenko at gmail dot com

fix the setup of Tx descriptors. the frame type and "no ack" bits were
initialized in ar5k_ar521*_setup_tx_desc() but cleared in
ar5k_ar521*_fill_tx_desc() by setting the segment length incorrectly.

From ath5k via Nick Kossifidis (mickflemm at gmail dot com)*

---snip---
The missing no_ack in broadcast frames caused them to be retried up to
the retry_limit(1+4=5 transmissions by default).
---snap---

* it was explicitly mentioned that it is ISC-licensed now

``it's'' -> ``its'' when the grammar gods require this change.

add some HAL functions (again) inspired by the changes from the
madwifi-old-openhal project. this was backed out the last time because
it was too close to the release and i didn't respect the ABI/API lock.

do not use bcopy to set the mac address registers. this diff should
fix the "hal" on big endian systems.

thanks to the madwifi-old-openhal people

back out, reyk is not paying attention

Add some HAL functions.

From madwifi-old-openhal via Nick Kossifidis (mickflemm at gmail dot com)

Bump copyright while I'm here.

Consistently spell FALLTHROUGH to appease lint.
ok kettenis@ cloder@ tom@ henning@

Update my e-mail address in the copyright statement, no binary changes.

KNF

fix a bug in Tx descriptor processing.

found by David Malone

don't attach unsupported radio chipsets

based on a diff by jsg@

ansi, no binary change.
ok reyk@

catched->caught

reduce stack usage

spotted out by jsg@

remove non-constant array initializers which is a GNU extension.
suggested by anil some time ago.

remove some misplaced htole32s

ath_hal_rx_monitor should be used for link state monitoring instead of
setting the promisc mode. rename it to ath_hal_rx_signal and remove
the incorrect promisc flag. this should fix some problems with ath
running in forwarding mode.

sync power management settings. this should fix some unrecoverable
sleep mode timeouts for ar5210, ar5211 and ar5212.

the ar5210 is silicon version 0, not 1.

fix silicon version number reporting. ie. ath0: mac 80.6 -> ath0: mac 5.6.

beautify the code by renaming HAL functions with capitalized words (i
always wanted to do that). this breaks HAL compatibility but porting
should be easy, have a look at athvar.h. no functional changes.

move reset of HAL descriptors to ath.c

bye bye unportable bitfields

pad keys with zeros to a supported key length. this is bad but seems
to be the default WEP behaviour for string-based keys and allows
commands like "ifconfig ath0 nwkey yuck" to work...

some fixes for wep and reset.

btw.: this is a commit using an ar5212 in 11b mode...

ok deraadt@, krw@

ignore bmiss interrupts for now.

fix beacon handling

a new year and a new, less restrictive license. because it has to be
free.

go ahead miod@

knf

cosmetic changes, fix regdomain code and beacon handling

fix an invalid condition which returns on an invalid beacon state.

fix regulation domain handling

sync with the latest work and add some stuff needed by the upcoming ar5211/ar5212
support. some further cleanups and changes will follow.

store number of supported gpio pins

some fixes for beeing compatible with gcc3 as noticed by some people.
thanks to Joerg Sonnenberger and Nathan Binkert.

use __packed instead of __attribute__ ((__packed__)) (ok miod@), check
possible NULL mac in the keycache function.

knf, use positive error values (suggested by miod@)

spacing, beautify dmesg output

removed some printfs

import of a free hal part for the ath driver as a replacement for the
binary-only hal module found in FreeBSD and NetBSD. OpenBSD's approach
is based on reverse engineering because it is _not_ possible to
include a non-free and binary-only piece of software in a 100% free
operating system. it still lacks some features found in the "official" hal
module but this will be done very soon with a help by a lot of
contributors - because it's free.

ok deraadt@