- 11 Nov, 2011 40 commits
-
-
Arend van Spriel authored
Instead of dealing with sk_buff prev pointers the queue functions now make use of the sk_buff_head functions provided by the kernel. Reported-by:
Johannes Berg <johannes@sipsolutions.net> Reviewed-by:
Alwin Beukers <alwin@broadcom.com> Reviewed-by:
Pieter-Paul Giesberts <pieterpg@broadcom.com> Signed-off-by:
Arend van Spriel <arend@broadcom.com> Signed-off-by:
John W. Linville <linville@tuxdriver.com>
-
Arend van Spriel authored
The function brcmu_pkttotlen calculates the total length of a sk_buff chain following the next pointer. In brcmsmac this is not needed as in each place where it was used the provided sk_buff had a NULL pointer as next field value. Reviewed-by:
-
Franky Lin authored
sdio_chip.c is dedicated to sdio bus chip. No need to retrieve id for bus core Reviewed-by:
Franky Lin <frankyl@broadcom.com> Signed-off-by:
-
Franky Lin authored
This patch is part of the series of adding new backplane support Reviewed-by:
-
Franky Lin authored
This patch is part of the series of adding new backplane support Reviewed-by:
-
Franky Lin authored
This patch is part of the series adding new backplane support Reviewed-by:
-
Franky Lin authored
New type of backplane interconnect support is needed for bcm4330 Reviewed-by:
-
Alwin Beukers authored
Reported-by:
-
Arend van Spriel authored
The macro definition D11_PHY_HDR_LEN was defined in d11.h and pub.h. It has been removed from pub.h. Reviewed-by:
Roland Vossen <rvossen@broadcom.com> Reviewed-by:
-
Arend van Spriel authored
The brcmsmac had own definitions for the access categories. The mac80211 header provides these as well as they are used in the conf_tx callback. As the definitions did not match the driver configured the tx parameters to the wrong queue. Reviewed-by:
-
Arend van Spriel authored
There were a couple of code segments left that were placed under a macro definition that was not applicable or not used in the brcmsmac driver. These pieces of code have been removed. Reviewed-by:
-
Arend van Spriel authored
This fixes issue when using OTP memory. It was introduced by following commit: commit 028f78d4 Author: Arend van Spriel <arend@broadcom.com> Date: Fri Oct 21 16:16:35 2011 +0200 brcm80211: smac: change buffer endianess convert function interface Signed-off-by:
-
Don Fry authored
The nvm_device_type is eeprom related and does not need to be part of the iwl_priv structure. Move it and eliminate access to the iwl_priv structure. Signed-off-by:
Don Fry <donald.h.fry@intel.com> Signed-off-by:
Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by:
-
Don Fry authored
Make the eeprom routines less dependent on the iwl_priv structure. Don't use the priv when bus structure is sufficient. Signed-off-by:
-
Venkataraman, Meenakshi authored
iwlwifi tries to avoid using antenna B in BT combo devices when BT is active. A bug in the rate-scaling algorithm was causing the combo device to never attempt MIMO rates. Fix the algorithm to opportunistically try MIMO rates when BT traffic is low. Signed-off-by:
Meenakshi Venkataraman <meenakshi.venkataraman@intel.com> Signed-off-by:
-
Emmanuel Grumbach authored
Users complain that the traffic gets stalled sometimes. This will allow easier debugging. Signed-off-by:
Emmanuel Grumbach <emmanuel.grumbach@intel.com> Signed-off-by:
-
Emmanuel Grumbach authored
This bit was used only once. Use IWL_DL_INFO instead. Signed-off-by:
-
Johannes Berg authored
Due to the P2P context always having the associated bit set when we got to this code, we were splitting up the remain-on-channel durations unconditionally. This isn't needed -- if the P2P context is in P2P device type mode it doesn't impose restrictions on timing to skip it in that case. Signed-off-by:
Johannes Berg <johannes.berg@intel.com> Signed-off-by:
-
Johannes Berg authored
The device expects TU but we get ms, so convert. Signed-off-by:
-
Emmanuel Grumbach authored
Warning introduced by c847474b7dfdda304d0d8ffcc5a9db546b1cb3e9 iwlwifi: check status before send command Signed-off-by:
-
Wey-Yi Guy authored
Check the status before sending host command, if any of the condition match, cancel the host command before queue Signed-off-by:
-
Wey-Yi Guy authored
"echo" host command has no data, set the length to 0 Signed-off-by:
-
Wey-Yi Guy authored
The core suspend function is part of agn, iwl_mac80211 should only handle mac80211 I/F operations. Signed-off-by:
-
Wey-Yi Guy authored
The station related mac80211 callback functions should belong to _mac80211 Signed-off-by:
-
Wey-Yi Guy authored
iwlagn_mac_hw_scan should belong to _mac80211 callback. Signed-off-by:
-
Wey-Yi Guy authored
Move more mac80211 related functions to _mac80211 file Signed-off-by:
-
Johannes Berg authored
In P2P device mode, the device needs to have valid QoS parameters. We currently have those because we program parameters from any virtual interface into all contexts, but not only do we want to get rid of this -- it is also unpredictable since on the BSS context we might have any parameters, and there it might even be programmed for HT. Explicitly program default QoS parameters into the PAN context for P2P (the defaults are 11g but with QoS disabled) to make device behaviour predictable. This also helps when in a follow-up patch we will use TX QoS parameters from mac80211 only for the context they were meant for -- without this first that would completely break P2P device discovery. Signed-off-by:
-
Johannes Berg authored
Eliad added the ability to configure AC parameters per virtual interface; make use of this in iwlwifi and set the parameters in the right context. Since storage and uploading to the device is already per context, this is sufficient. Signed-off-by:
-
Wey-Yi Guy authored
There are many mac80211 callback functions in the driver, as current implementation, those functions are scatter everywhere which is very difficult to find and maintain, move all the mac80211 callback functions into single file. There should be no functional changes Signed-off-by:
-
Don Fry authored
Relocate the ucode files and update relevant code. More code refactoring. Signed-off-by:
-
Don Fry authored
Move the knowledge of the ucode image to lower level routines. Also do not pass the iwl_priv structure lower than it needs to be. Signed-off-by:
-
Don Fry authored
The upper layers of the driver do not need to know where the ucode is stored. It already passes in an enum of which ucode to load. Let the lower layer routines select the ucode to load. Signed-off-by:
-
Don Fry authored
ucode_write_complete is used for ucode loading. Move it as part of restructuring work out of the priv structure. Signed-off-by:
-
Wey-Yi Guy authored
BG only SKUs are no longer supported by 2000 and 1x5 series. Remove it Signed-off-by:
-
Johannes Berg authored
Reinette reports a crash in iwl_reprogram_ap_sta(). The debugging shows: b1 16 mov $0x16,%cl *f3 a5 rep movsl %ds <-- trapping instruction:(%rsi),%es:(%rdi) which is a memcpy of 22 (0x16) words (movsl). this points to "priv->stations[sta_id].lq" being NULL since that is the memcpy() of that size here. The only way I see for this to happen is if we try to do some RXON reprogramming while connecting to an AP, after tx_sync() but before full setup, but that seems like something that might very well happen. Fix this by checking if the LQ is present and only then reprogramming it. Reported-by:
Reinette Chatre <reinette.chatre@intel.com> Signed-off-by:
-
Wey-Yi Guy authored
Check and report WARN only when its invalid Signed-off-by:
-
Wey-Yi Guy authored
Add two more SKUs for 6x05 series of device. First SKU has low 5GHz channels actives, the other SKU has high 5GHz channels actives. Signed-off-by:
-
Todd Previte authored
When a station is associated with an AP and the RF_KILL switch is engaged, numerous error messages were sent to the system log. The error messages were the result of the failure(s) of the various submodules to perform their tasks after the radios were disabled. To resolve this situation, the messages were modified to use a new macro, IWL_DEBUG_QUIET_RFKILL. This macro allows for the RF_KILL error messages to be sent to the log provided that IWL_DEBUG is true and IWL_DL_RADIO is '1'. For all other cases, the error messages resulting from an RFKILL event will not be sent to the system log. Messages logged because of an RFKILL will be tagged with the prefix '(RFKILL)' to clarify the cause of the error. Signed-off-by:
Todd Previte <toddX.a.previte@intel.com> Signed-off-by:
-
Johannes Berg authored
Whether to use NoA or not is entire controlled by the uCode right now, and it also adds the attribute to beacons. We do need to add it to probe responses in the driver though. Keep track of the NoA notification from the uCode and add the data to probe responses when such are transmitted. Use RCU to manage the lifetime. Signed-off-by:
-
Patrick Kelle authored
Signed-off-by:
Patrick Kelle <patrick.kelle81@gmail.com> Signed-off-by:
-
