12 Sep, 2008
40 commits
-
There isn't really a good reason to have the LED
configuration options selectable per driver, lets
make it default 'y' and make it depend on the
NEW_LEDS and LEDS_CLASS interface.Signed-off-by: Ivo van Doorn
-
RFKILL should be enabled for _all_ hardware whether
or not they feature a rfkill button or not.Remove driver specific RFKILL configuration options
and make the rt2x00lib version depend on CONFIG_RFKILL
and defaulting to 'y' to make sure it will always
be enabled when RFKILL was enabled.This also fixes some bugs where RFKILL wasn't initialized
and didn't respond to RFKILL key presses.Signed-off-by: Ivo van Doorn
Signed-off-by: John W. Linville -
Cleans up the code a bit and prepares for the next patch
that will use the function elsewhere.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
deauth and disassoc frames are completely identical so there's
little point in having two functions to send them rather than
one that gets a parameter. This same a bit of code size.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This reorders all frame sending functions to be at the top of the
file. When reading the file, I tend to be looking at either the
frame code or the state machine, and having them mixed in the file
is confusing. When all frame sending is at the top the remainder
of the file is more readable, in my opinion.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Ever since we refactored beaconing to not be controlled by a
fake queue this parameter to ieee80211_sta_def_wmm_params
has been unused.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
ieee80211_sta_start_scan() can very well take a non-NULL
ssid pointer with a zero ssid_len.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
When a scan is requested for non-STA interfaces, we simply fire
off a scan, but for STA interfaces we shouldn't because they
could be in the middle of an association. This clarifies the
corresponding code.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Now that I've created ht.c, I can move the aggregation
code from main.c into it.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Some of the HT code in mlme.c is misplaced:
* constants/definitions belong to the ieee80211.h header
* code being used in other modes as well shouldn't be thereSigned-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This moves all the BSS list handling out of mlme.c to scan.c,
no further changes except fixing kzalloc/atomic_inc/atomic_inc
to kzalloc/atomic_set(2).Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This patch refactors some code and moves the scan RX function
to scan.c. More importantly, however, it changes it so that the
MLME's beacon/probe_resp functions aren't invoked when scanning
so that we can remove a "if (scanning)" conditions from two
places.There's a very slight behavioural change in this patch: now,
when scanning, IBSS and mesh aren't updated even on the same
channel.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Here's a first patch to move some code from mlme.c to a
new file called scan.c. The end result will hopefully be
a more manageable mlme.c.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
I'm surprised nobody complained about these before. What a waste.
Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Since we now use sdata pointers most of the time, using a netdev
pointer here is somewhat artificial, use an sdata pointer instead.
Replace a netdev-prefix in a few messages by a wiphy-prefix.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This reorders the mlme code a bit so we don't need all the forward
function declarations. It also removes the ERP_INFO_USE_PROTECTION
define that is unused, but otherwise contains no real changes.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
During testing of the disassociation fixes, Tomas noticed that it
was possible to run into a situation where you'd suddenly get a
few "wlan0: dropped frame to (unauthorized port)" messages
and I found this to be due to the AP's sta_info having been
removed but netif_carrier_off not having removed/stopped traffic
yet. To avoid that, stop the queue for the interface (and avoid
bringing them up when another vif scans when they weren't up.)Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This patch moves disassociation code from ieee80211_set_associated
to ieee80211_set_disassoc. To reduce code duplication, it introduces
the ieee80211_sta_send_apinfo function. Additionally, it fixes a lapse
where BSS_CHANGED_HT wasn't set when notifying the driver of changes
due to disassociation.Signed-off-by: Tomas Winkler
Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This patch makes the MLME cleanly disassociate from the current BSS
when leaving it for a new one. This is not just nicer to the old AP
(we're leaving it, might as well tell it!) but also required for some
drivers that keep track of the station we're associated with, they'd
get confused because they'd think we are associated with two APs.Signed-off-by: Ron Rindjunsky
Signed-off-by: Tomas Winkler
Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This patch restructure the flow of disassociation and deauthentication
flows to be consistent under all circumstances.
It ensures that BA session is treated down before deauthentication or disassociation,
adds the removal of the obsolete sta form station table and fixes a related bug (sta_info_destroy
without sta_info_unlink) in ieee80211_associated()
and reduce some code duplicationSigned-off-by: Ron Rindjunsky
Signed-off-by: Tomas Winkler
Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This reverts:
b93ce437eba7e0232683326f30d9d1167a872fad
rt2x00: Fix the beacon length bugThe workaround is no longer required since it has been
correctly fixed in rt2x00usb now.Signed-off-by: Ivo van Doorn
Signed-off-by: John W. Linville -
Since IE parsing is required for the mlme and mesh code, it's
not a static function anyway, and it's much better to have it
in util rather than the overly large mlme.cSigned-off-by: Johannes Berg
Signed-off-by: John W. Linville -
This says chan_switch.action_code but really means
measurement.action_code, of course the actual offset in
the frame is the same, it's just harder to understand
this way.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
The action frame length checks are one too small, there's not just
an action code as the comment makes you believe, there's a category
code too, and the category code is required in each action frame
(hence part of IEEE80211_MIN_ACTION_SIZE).Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
When we receive information about a BSS we check at some point
whether or not we think we're allowed to use the channel it is
on, but we do that fairly late. I don't think we should do it
that late, so do it earlier to avoid doing IBSS/mesh stuff on
that channel and then getting confused because it's disabled.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
The conf_tx callback currently needs to be atomic, this requirement
is just because it can be called from scanning. This rearranges it
slightly to only update while not scanning (which is fine, we'll be
getting beacons when associated) and thus removes the atomic
requirement.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
The if itself doesn't need to be protected, so move in the RCU
locking to avoid doing anything at all when the condition isn't
true.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Recent changes in the specifications have improved the performance
of the BCM4306/2 devices that use b43legacy as the driver. These
"errors" in the specs have been present from the very first implementation
of bcm43xx.Signed-off-by: Ehud Gavron
Tested-by: Larry Finger
Signed-off-by: John W. Linville -
A coding error present since b43legacy was incorporated into the
kernel has prevented the driver from using the rate-setting mechanism
of mac80211. The driver has been forced to remain at a 1 Mb/s rate.Signed-off-by: Larry Finger
Cc: Stable [2.6.26], [2.6.25]
Signed-off-by: John W. Linville -
We don't need the workqueue anymore, as we can now sleep in the callback.
Signed-off-by: Michael Buesch
Signed-off-by: John W. Linville -
It causes compile errors on m68k and it is not needed.
Remove it.Signed-off-by: Michael Buesch
Signed-off-by: John W. Linville -
This fixes the initialization of the default QoS parameters.
This got broken by "wireless: fix warnings from QoS patch".Reported-by: Lorenzo Nava
Signed-off-by: Michael Buesch
Signed-off-by: John W. Linville -
This patch brings the 5GHz Phy in any prism54 devices (of course, only those who have one) to life.
Signed-off-by: Christian Lamparter
Signed-off-by: John W. Linville -
I hope this patch is enough to cover at least the basic requirements of IEEE 802.11h's TPC.
Signed-off-by: Christian Lamparter
Signed-off-by: John W. Linville -
The firmware can provide lots of useful statistics about noise floor,
mac time and lots of numbers about successful transfers and dropped
frames.Signed-off-by: Christian Lamparter
Signed-off-by: John W. Linville -
This patch adds new filters settings to make the card more useful in monitor mode.
Signed-off-by: Christian Lamparter
Signed-off-by: John W. Linville -
tcpdump:
02:15:42.874518 61112184us tsft 48.0 Mb/s 2437 MHz (0x0480) antenna 1 [0x0000000e] CF +QoS Data IV
02:15:42.874557 >>>4356079526us<<< tsft 24.0 Mb/s 2437 MHz (0x0480) antenna 1 [0x0000000e] Acknowledgment
02:15:42.976844 61214513us tsft 1.0 Mb/s 2437 MHz (0x0480) antenna 0 [0x0000000e] Beaconas one can see on the huge jump, it's very plausible that firmware does not report the
full 64-bit mac time, just the lower 32bit and some kinds of flags...
Therefore if we want a useful timestamp we have to emulate the high bits.Signed-off-by: Christian Lamparter
Signed-off-by: John W. Linville -
The patch fixes compile warning for ‘iwl4965_hw_channel_switch’ defined
but not used.Signed-off-by: Zhu Yi
Signed-off-by: John W. Linville -
This patch cleans up iwlwifi by removing uneeded declarations and removing
uneeded symbol export reducing the namespace pollution. It also fixes some
typos in comments.Signed-off-by: Emmanuel Grumbach
Signed-off-by: Zhu Yi
Signed-off-by: John W. Linville -
This patch makes usage of the results from disconnected antenna alg to
know how many antennas are connected.It also synchronizes between the chain noise alg and the W/A that
disables power management during association. All the antennas must be
enables during the chain noise algorithm. Hence, power management is
restored only after the completion of the algorithm.In the future, we will need to update the AP that we don't support MIMO
if there is only one antenna connected. We also need to update the rate
scaling algorithm.Signed-off-by: Emmanuel Grumbach
Signed-off-by: Tomas Winkler
Signed-off-by: Zhu Yi
Signed-off-by: John W. Linville
