- 10 Jul, 2009 40 commits
-
-
Johannes Berg authored
When the auth algorithm is rejected, but we don't have another one to try, we will eventually retry but that isn't useful -- we'll then do it again and again until we eventually give up. Instead, we should let the SME know and go into disabled state. The same applies for situations where the AP rejects with any other status code. Additionally, when trying the next auth algorithm, we should reset the auth_tries so that just a single lost frame doesn't lead to us giving up on the third auth algorithm. Signed-off-by:
Johannes Berg <johannes@sipsolutions.net> Signed-off-by:
John W. Linville <linville@tuxdriver.com>
-
Johannes Berg authored
This variable isn't necessary -- the wext code keeps track of the BSSID itself, and otherwise we have current_bss. Signed-off-by:
-
Johannes Berg authored
Instead of hardcoding GFP_ATOMIC everywhere, add a new function parameter that gets the flags from the caller. Obviously then I need to update all callers (all of them in mac80211), and it turns out that now it's ok to use GFP_KERNEL in almost all places. Signed-off-by:
-
Johannes Berg authored
I don't like the 'extern' keyword much when it's not necessary, it makes lines rather long. Signed-off-by:
-
Johannes Berg authored
Move a break statement to the correct place _after_ the #endif, otherwise w/o WIRELESS_EXT things break badly. Also, while touching this code, do a cleanup and assign dev->ieee80211_ptr to a new variable. Signed-off-by:
-
Johannes Berg authored
The way I initially thought we could do wireless extensions is by making all the compat code in cfg80211 be independent of CONFIG_WIRELESS_EXT, but this is turning out to not be feasible. Therefore, fix the Kconfig help text and make the option default to yes, so people won't get a nasty surprise when mac80211 will get rid of its 'select WIRELESS_EXT' any time now. Signed-off-by:
-
Johannes Berg authored
The key todo lock can be taken from different locks that require it to be _bh to avoid lock inversion due to (soft)irqs. This should fix the two problems reported by Bob and Gabor: http://mid.gmane.org/20090619113049.GB18956@hash.localnet http://mid.gmane.org/4A3FA376.8020307@openwrt.orgSigned-off-by:
-
Johannes Berg authored
Currently, wext drivers cannot return NULL for stats even though that would make the ioctl return -EOPNOTSUPP because that would mean they are no longer listed in /proc/net/wireless. This patch changes the wext core's behaviour to list them if they have any wireless_handlers, but only show their stats when available, so that drivers can start returning NULL if stats are currently not available, reducing confusion for e.g. IBSS. Signed-off-by:
-
John W. Linville authored
Signed-off-by: -
Gabor Juhos authored
This helper only clears the SC_OP_WAIT_FOR_{BEACON,CAB} flags. Remove it and clear these flags directly in the approptiate places instead. Changes-licensed-under: ISC Signed-off-by:Gabor Juhos <juhosg@openwrt.org> Signed-off-by:
-
Gabor Juhos authored
All SC_OP_WAIT_* flags will be cleared in 'ath9k_conf' when PS mode is disabled, so we don't have to clear it here. Changes-licensed-under: ISC Signed-off-by:
-
Johannes Berg authored
Instead of having mac80211 do it itself. Signed-off-by:
-
David Kilroy authored
Signed-off-by:
David Kilroy <kilroyd@googlemail.com> Signed-off-by:
-
David Kilroy authored
Signed-off-by:
-
David Kilroy authored
This removes the custom scan cache used by orinoco. We also have to avoid calling cfg80211_scan_done from the hard interrupt, so we offload the entirety of scan processing to a workqueue. This may behave strangely if you start scanning just prior to suspending... Signed-off-by:
-
David Kilroy authored
Signed-off-by:
-
David Kilroy authored
This allows changes to be commited from cfg80211 functions. Signed-off-by:
-
David Kilroy authored
Each device does almost exactly the same things on suspend and resume when upping and downing the interface. So move this logic into a common routine. Signed-off-by:
-
David Kilroy authored
... instead of relying on the net_device fields. Signed-off-by:
-
David Kilroy authored
With the move to cfg80211 it's nice to keep the hardware operations distinct from the interface, even though we can only support a single interface. This also means the driver resembles other cfg80211 drivers. Signed-off-by:
-
David Kilroy authored
The firmware download code has been in a couple of releases, without any significant issues reported in this code. Convert to use pr_debug, so the messages can be recoverred by defining DEBUG. Signed-off-by:
-
David Kilroy authored
Initialise and register a wiphy. Store the orinoco_private structure in the new wiphy, and use the net_device private area to store the wireless_dev. This results in a change to the way we navigate from a net_device to the driver private orinoco_private, which we encapsulate in the inline function ndev_priv. Most of the remaining calls to netdev_priv are thus replaced by ndev_priv. We can immediately rely on cfg80211 to handle SIOCGIWNAME, so orinoco_ioctl_getname is removed. Signed-off-by:
-
David Kilroy authored
hw.h does not include hermes.h, and none of the other functions requires types from that file. Also hermes_t is a (discouraged) typedef so we can't add a forward declaration. Therefore change this function to use orinoco_private. Signed-off-by:
-
David Kilroy authored
Initialise the orinoco driver before registerring with netdev, which will help when we get to cfg80211... Signed-off-by:
-
David Kilroy authored
Move away from using net_device as the main structure in orinoco function calls. Use orinoco_private instead. This makes more sense when we move to cfg80211, and we get wiphys as well. Signed-off-by:
-
David Kilroy authored
We should be able to call these routines before we register with netdev, so avoid printks using the netdev name. Signed-off-by:
-
David Kilroy authored
This allows us to use determine_fw_capabilities, orinoco_hw_read_card_setting and orinoco_hw_allocate_fid prior to netdev registration. Since dev_dbg only prints if DEBUG is defined (or dynamic debug is enabled), move a couple of the more useful prints up to info. Signed-off-by:
-
David Kilroy authored
This is part of refactorring the initialisation code so that we can load the firmware before registerring with netdev. Signed-off-by:
-
David Kilroy authored
This is part of refactorring the initialisation code so that we can load the firmware before registerring with netdev. Signed-off-by:
-
David Kilroy authored
This is part of refactorring the initialisation code so that we can load the firmware before registerring with netdev. Signed-off-by:
-
David Kilroy authored
Signed-off-by:
-
David Kilroy authored
Only set the sizes for WEP40 and WEP104. Signed-off-by:
-
David Kilroy authored
Signed-off-by:
-
David Kilroy authored
Signed-off-by:
-
David Kilroy authored
Signed-off-by:
-
Johannes Berg authored
We need to clean up the skb before we can copy it, this is required for proper operation since the socket it is still attached to could potentially live in a different network namespace or so. Signed-off-by:
-
Hin-Tak Leung authored
In the past ids were inserted in no particular order and it has become hard to see what ids are already included or not. This patch sorts the ids by numerical value and has no functional effect. Signed-off-by:
Hin-Tak Leung <htl10@users.sourceforge.net> Signed-off-by:
-
Johannes Berg authored
It isn't very useful to scan the same channel more than once during a given scan, and some hardware (notably iwlwifi) can only scan a limited number of channels at a time. To prevent any overflows, simply disallow scanning any channel multiple times in a given scan command. This is a small change in the userspace ABI, but the only user, wpa_supplicant, never asks for a scan with the same frequency listed twice. Signed-off-by:
-
Johannes Berg authored
We had code for a number of files, that we didn't publish in debugfs, fix that. Also make the agg_status file layout more readable and add more information to it. Signed-off-by:
-
Johannes Berg authored
Within mac80211, we often need to copy the rx status into skb->cb. This is wasteful, as drivers could be building it in there to start with. This patch changes the API so that drivers are expected to pass the RX status in skb->cb, now accessible as IEEE80211_SKB_RXCB(skb). It also updates all drivers to pass the rx status in there, but only by making them memcpy() it into place before the call to the receive function (ieee80211_rx(_irqsafe)). Each driver can now be optimised on its own schedule. Signed-off-by:
-
