- 14 Aug, 2009 40 commits
-
-
Andrey Yurovsky authored
On locally originated traffic, we refresh active paths after a timeout. The decision to do this was using the wrong sign and therefore the refresh timer was triggered for every frame. Signed-off-by:
Andrey Yurovsky <andrey@cozybit.com> Signed-off-by:
Javier Cardona <javier@cozybit.com> Signed-off-by:
John W. Linville <linville@tuxdriver.com>
-
Javier Cardona authored
Also, fix typo in comment. Signed-off-by:
-
Gábor Stefanik authored
The rev2+ BB init spec has changed behind us, and thus the code is no longer up to date. Update the code to match the current specs. Also implement save/restore dig filt state, as required by the new specification. Signed-off-by:
Gábor Stefanik <netrolller.3d@gmail.com> Signed-off-by:
-
Gábor Stefanik authored
Implement LP-PHY baseband table init for all revisions. Signed-off-by:
-
Johannes Berg authored
It's really easier to read if it's not indented as much, so invert the condition and rearrange the code so the smaller chunk is indented instead. Signed-off-by:
Johannes Berg <johannes@sipsolutions.net> Signed-off-by:
-
Johannes Berg authored
There's no need to mask the variable with 0xFFF0 since we ever only use it as a u16 and the lowest four bits can't ever be non-zero. The compiler cannot infer the latter, and therefore has to emit code to do the masking. Signed-off-by:
-
Johannes Berg authored
When encryption is used, the number of bytes sent to the peer increases by the IV and ICV. This is accounted if software encryption is used, but not if the devices does hardware encryption. To make the numbers comparable, never account for that overhead. Signed-off-by:
-
Luciano Coelho authored
Add maintainer information section for the wl1271 wireless driver and fix the information for wl1271. Signed-off-by:
Luciano Coelho <luciano.coelho@nokia.com> Acked-by:
Kalle Valo <kalle.valo@nokia.com> Signed-off-by:
-
Nick Kossifidis authored
* Get rid of calibration timer, instead use a software interrupt to schedule the calibration tasklet. a) We don't need a timer for this, there is no need for accuracy even with round_jiffies i think this is a waste of resources. Also we don't need to run calibration if we are idle (no interrupts). b) When we add ANI support we 'll just extend the poll function and calibration tasklet and handle all periodic phy calibration on one place (much cleaner). c) Having calibration on a tasklet is better since during calibration we can't transmit or receive (antennas are detached to measure noise floor), previously calibration could run in parallel with tx/rx and interfere (packet loss). v2: kill tasklet on stop_hw, stop/wake queues v3: use time_is_before_eq_jiffies to compare timestamp with current time Signed-off-by:Nick Kossifidis <mickflemm@gmail.com> Signed-off-by:
-
Nick Kossifidis authored
* During attach preserve pcicfg bits when enabling pci core sw retry fix. Signed-off-by:
-
Nick Kossifidis authored
* Don't put chip to full sleep because there are problems during wakeup. Instead hold MAC/Baseband on warm reset state via a new function ath5k_hw_on_hold. * Minor cleanups Signed-off-by:
Ben Greear <greearb@candelatech.com> Tested-by:
Johannes Stezenbach <js@sig21.net> Signed-off-by:
-
Nick Kossifidis authored
* Set correct xpd curve indices for high/low gain curves during rfbuffer setup on RF5112B with both calibration curves available. * Don't return zero min power when we have the same pcdac value twice because it breaks interpolation. Instead return the right x barrier as we do when we have equal power levels for 2 different pcdac values. Signed-off-by:
Bob Copeland <me@bobcopeland.com> Signed-off-by:
-
Nick Kossifidis authored
* Read PCI-E infos offset from EEPROM and if it points to serdes section (0x40), enable serdes programming (further tweaking of serdes values during attach). This follows Legacy and Sam's HAL sources. Signed-off-by:
-
Igor Perminov authored
When beacon is being updated to refresh TIM (AP mode), beacon frames are de-synchronizing (i.e. two neighbor beacon frames - before and after update - are being transmitted with a wrong time interval). That is because xxx_write_beacon should disable beacon generation only while beacon data are being uploaded to the device, but it should not disable the beacon clock. Signed-off-by:
Igor Perminov <igor.perminov@inbox.ru> Signed-off-by:
Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by:
-
Gábor Stefanik authored
The v2+ radio init (B2063) is now complete, modulo BCM4325 support. Signed-off-by:
-
Michael Buesch authored
Add some TX desc flags docs. Signed-off-by:
Michael Buesch <mb@bu3sch.de> Signed-off-by:
-
Johannes Berg authored
When calling into the wext code from the NETDEV_UP notifier, we need to hold the devlist_mtx mutex as the wext code ends up calling into channel checks. Reported-by:
Kalle Valo <kalle.valo@iki.fi> Signed-off-by:
-
Luis R. Rodriguez authored
In the the patch: ath9k: use new FIF_PSPOLL configure filter I forgot to add the new FIF_PSPOLL to the supported mask. Without this we mask out the FIF_PSPOLL flag therefore we'd ignore it from mac80211. Cc: Jouni Malinen <jouni.malinen@atheros.com> Signed-off-by:
Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by:
-
Pat Erley authored
This removes the max_bandwidth attribute. It is only ever written to, and is duplicated by max_bandwidth_khz in the regulatory code. Signed-off-by:
Pat Erley <pat-lkml@erley.org> Acked-by:
-
Igor Perminov authored
This patch implements FIF_PSPOLL filter flag support in rt2x00 drivers, which has been introduced in mac80211 (see http://marc.info/?l=linux-wireless&m=124897986330807&w=2). Signed-off-by:
-
Igor Perminov authored
The patch "Implement set_tim callback for all drivers" can cause kernel oops in rt73usb_write_beacon. The oops is caused by one of the following race conditions: * In case of two near calls to set_tim: rt2x00lib_beacondone_iter is cleaning the beacon skb, whereas rt73usb_write_beacon is still using it. * In case of two near updates of beacon: first as the result of set_tim and second as the result of a call from an application (e.g. hostapd). This patch fixes the race condition by rearranging the update logic and guarding rt2x00_intf->beacon->skb with a mutex. Signed-off-by:
-
Lars Ericsson authored
Small changes in signal level was not detected up by the MOVING_AVERAGE() due to a rounding error, using 'int' type. rt2x00lib_antenna_diversity_eval: rssi: -62 -62 -62 -62 -62 -62 -62 -62 -62 -62 -62 -62 -62 -62 -62 rssi_avg: -57 -57 -57 -57 -57 -57 -57 -57 -57 -57 -57 -57 -57 -57 -57 The signal level reported back could be significantly (5dBm) different from the actual value. A level +3dBm is the same as double the AP output power. Signed-off-by:
Lars Ericsson <Lars_Ericsson@telia.com> Signed-off-by:
-
Lars Ericsson authored
Antanna quality statistics is not handled correctly, which leads to software diversity being shutdown completly. The main problem is that during antenna diversity statistics can be reset resulting in loosing the signal strength just before evaluation. rssi history is not updated correctly leading to incorrect comparison material for basing antenna switching on. Signed-off-by:
-
Ivo van Doorn authored
Some hardware require the ieee80211 header to be aligned to a 4-byte boundary before mapping it to the DMA. Otherwise some frames (like beacons) will not be send out correctly by the device. Signed-off-by:
-
Ivo van Doorn authored
Check the IEEE80211_TX_CTL_MORE_FRAMES flag to help determining if the DMA queue should be kicked. At the moment this is combined with the ieee80211_has_morefrags() but we might remove that later. Signed-off-by:
-
Ivo van Doorn authored
In the config() callback function the fields radio_enabled and the change flag IEEE80211_CONF_CHANGE_RADIO_ENABLED have been deprecated. This removes the usage of those fields by improving antenna change detection in the antenna configuration function. Signed-off-by:
-
Julia Lawall authored
Elsewhere, the tqi_type field is compared to constants having a name beginning with AR5K_TX_QUEUE, rather than AR5K_TX_QUEUE_ID. I have thus converted AR5K_TX_QUEUE_ID_CAB to AR5K_TX_QUEUE_CAB. This does, however, change the value, so perhaps something else was wanted. Signed-off-by:
Julia Lawall <julia@diku.dk> Acked-by:
-
Johannes Berg authored
"cfg80211: validate channel settings across interfaces" contained a locking bug -- in the managed-mode SIWFREQ call it would end up running into a lock recursion. This fixes it by not checking that particular interface for a channel that it needs to stay on, which is as it should be as that's the interface we're setting the channel for. Reported-by:
Reinette Chatre <reinette.chatre@intel.com> Reported-by:
-
Johannes Berg authored
Depending on required latency requested by pm_qos (via mac80211) we can automatically adjust the sleep state. Also, mac80211 has a user-visible dynamic sleep feature where we are supposed to stay awake after sending/receiving frames to better receive response frames to our packets, this can be integrated into the sleep command. Currently, and this patch doesn't change that yet, we default to using sleep level 1 if PS is enabled. With a module parameter to iwlcore, automatic adjustment to changing network latency requirements can be enabled -- this isn't yet the default due to requiring more testing. The goal is to enable automatic adjustment and then go into the deepest possible sleep state possible depending on the networking latency requirements. This patch does, however, enable IEEE80211_HW_SUPPORTS_DYNAMIC_PS to avoid the double-timer (one in software and one in the device) when transmitting -- the exact timeout may be ignored but that is not of big concern. Note also that we keep the hard-coded power indices around for thermal throttling -- the specification of that calls for using the specified power levels. Those can also be selected in debugfs to allow easier testing of such parameters. Signed-off-by:
-
Wey-Yi Guy authored
Do not send CT KILL config command twice and correct critical temperature informatiom in dmesg Signed-off-by:
Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by:
-
Reinette Chatre authored
When compiling without CONFIG_IWLWIFI_DEBUGFS there is a missing iwl_update_stats symbol. This is fixed by making this function an inline in the case when CONFIG_IWLWIFI_DEBUGFS is not set due to the hot path in which it is used. Signed-off-by:
-
Johannes Berg authored
Some of the thermal throttle data structures and code are really very intermingled with the sleep (power) control code. They really do belong together in a way since the thermal throttle code uses powersaving to achieve its goal, but it's making it hard to work on the powersave code. Split this up to make that easier. I've also changed the antenna defines to an enum and used the same enum for RX and TX. Signed-off-by:
-
Wey-Yi Guy authored
Modify the power update function, when driver fail to set the power, it should not continue move forward and try to change the rx chain configuration. Signed-off-by:
-
Wey-Yi Guy authored
For legacy thermal throttling, set the new Thermal Throttling state and change power index when thermal throttling manager detects temperature changed. The current implementation sets the state to the previous Thermal Throttling state, which causes system to enter wrong power index. The worse case, it will trying to set the lower power index when device reach critical temperature, it will cuase issue for both system and the device. Signed-off-by:
-
Wey-Yi Guy authored
Move all the thermal throttling functions to background task to make sure do not change power and rx chain in interrupt handler. Signed-off-by:
-
Reinette Chatre authored
commit "iwlwifi: uCode Alive notification with timeout" introduced a more reliable mechanism for ucode loading. Unfortunately we hit a problem with it frequently enough to make a 4965 unusable. The problem can be seen in debug log below. What this code attempts is to set runtime ucode up to load, start a timer to wait for the alive response from runtime ucode, and if it times out it tries again. As can be seen below we receive the alive response and wake the waiting task _before_ the tasks starts waiting. The task thus times out as the alive response is not received while it is waiting for it and it restarts the device. This starts the cycle all over again. [29739.000819] ieee80211 phy0: U iwl_mac_start enter [29739.005751] ieee80211 phy0: U iwl_prepare_card_hw iwl_prepare_card_hw enter [29739.012798] ieee80211 phy0: U iwl_set_hw_ready hardware ready [29739.057200] ieee80211 phy0: U iwl4965_load_bsm Begin load bsm [29739.063366] ieee80211 phy0: U iwl4965_verify_bsm Begin verify bsm [29739.072485] ieee80211 phy0: U iwl4965_verify_bsm BSM bootstrap uCode image OK [29739.079671] ieee80211 phy0: U iwl4965_load_bsm BSM write complete, poll 0 iterations [29739.257019] ieee80211 phy0: I iwl_rx_reply_alive Alive ucode status 0x00000001 revision 0x1 0x9 [29739.260964] ieee80211 phy0: I iwl_rx_reply_alive Initialization Alive received. [29739.260964] ieee80211 phy0: U __iwl_up iwlagn is coming up [29739.278571] ieee80211 phy0: U iwl_mac_start Start UP work done. [29739.284509] ieee80211 phy0: U iwlcore_verify_inst_sparse ucode inst image size is 788 [29739.292432] ieee80211 phy0: U iwlcore_verify_inst_sparse ucode inst image size is 10312 [29739.302004] ieee80211 phy0: U iwl_verify_ucode Initialize uCode is good in inst SRAM [29739.309746] ieee80211 phy0: U iwl4965_hw_get_temperature Running temperature calibration [29739.317833] ieee80211 phy0: U iwl4965_hw_get_temperature Calib values R[1-3]: -36 13522 -13496 R4: -2726 [29739.327337] ieee80211 phy0: U iwl4965_hw_get_temperature Calibrated temperature: 310K, 37C [29739.335598] ieee80211 phy0: U iwl4965_init_alive_start Initialization Alive received. [29739.343477] ieee80211 phy0: U iwl4965_set_ucode_ptrs Runtime uCode pointers are set. [29739.351283] ieee80211 phy0: I iwl_rx_reply_alive Alive ucode status 0x00000001 revision 0x1 0x0 [29739.355210] ieee80211 phy0: I iwl_rx_reply_alive Runtime Alive received. [29739.366731] iwlagn 0000:03:00.0: Runtime uCode already alive? Waiting for alive anyway [29743.284110] iwlagn 0000:03:00.0: START_ALIVE timeout after 4000ms. [29743.290337] ieee80211 phy0: U iwl_mac_add_interface enter: type 2 [29744.364089] iwlagn 0000:03:00.0: Runtime timeout after 5000ms [29744.370882] ieee80211 phy0: U iwl_alive_start Runtime Alive received. [29744.377347] ieee80211 phy0: U iwlcore_verify_inst_sparse ucode inst image size is 788 [29744.385287] ieee80211 phy0: U iwlcore_verify_inst_sparse ucode inst image size is 10312 [29744.393397] ieee80211 phy0: U iwlcore_verify_inst_sparse ucode inst image size is 94720 [29744.415835] ieee80211 phy0: U iwl_verify_ucode Runtime uCode is good in inst SRAM Signed-off-by:
-
Wey-Yi Guy authored
Display sensitivity and chain noise data to help understand the current environment and RF condition. The data is feeded by statistics notification and Beacon from uCode; then used by sensitivity calibration and chain noise calibration to determine how DSP should react to the environment changes Signed-off-by:
-
Wey-Yi Guy authored
Display statistics notification information The information break down into uCode_tx_stats uCode_rx_stats uCode_general_stats and can be found in /sys/kernel/debug/ieee80211/phy0/iwlagn/debug directory The statistic information display in debugFs is based on the last statistics notification from uCode; it might not reflect the current uCode activity. Using "watch" command to monitor the uCode activity should give up-to-date statistics provided by uCode. Signed-off-by: -
Wey-Yi Guy authored
Adding debugfs function to show current TxFifo/RxFifo read/write pointer, plus the current tx queue status (wake/stop) for both real and virtual queue. This is part of debug feature set to help debugging driver/uCode. use tx_queue and rx_queue in /sys/kernel/debug/ieee80211/phy0/iwlagn/debug directory to show the current read/write pointer for both TxFifo and RxFifo queue Signed-off-by:
-
Wey-Yi Guy authored
Break down the traffic type and counter for both Tx and Rx. Enhance the tx_statistics and rx_statistics debugfs function and move to /sys/kernel/debug/ieee80211/phy0/iwlagn/debug directory to help better debugging both driver and uCode related problems. Signed-off-by:
-
