- 20 Mar, 2013 4 commits
-
-
Emmanuel Grumbach authored
The masks were wrong. They should be 0xffffffff when SCO, HID or SNIFF profiles are used. They should be 0xffff0000 in any other case (default) to get a bit more throughput when the BT profile allows for it. Fix a debug print on the way. Signed-off-by:
Emmanuel Grumbach <emmanuel.grumbach@intel.com> Signed-off-by:
Johannes Berg <johannes.berg@intel.com>
-
Ilan Peer authored
Mostly for debugging purposes Signed-off-by:
Ilan Peer <ilan.peer@intel.com> Reviewed-by:
-
Johannes Berg authored
In AP mode, don't attempt to program GTKs into the device, they're used for TX only so not needed and programming them causes error messages. Also, in this case and if key programming fails, avoid trying to remove the key that isn't present later. Reviewed-by:
-
Emmanuel Grumbach authored
The code removed in this patch was used for bring up on older NICs. No MVM capable fw will ever be released for older NICs, so remove that code. Signed-off-by:
-
- 06 Mar, 2013 32 commits
-
-
Emmanuel Grumbach authored
The SCD byte count layout is decided by the configuration done in fw, it is then logical to export it as a TLV flag and not per HW SKU. Signed-off-by:
-
Amnon Paz authored
Fix a bug in writing to indirect (periphery) registers; although writes seem successful the data is not written to the desired address). Also fix address mask for HBUS_TARG_PRPH_RADDR and HBUS_TARG_PRPH_WADDR registers. Signed-off-by:
Amnon Paz <amnonX.paz@intel.com> Reviewed-by:
-
Emmanuel Grumbach authored
This allows to test fw restart flow. The hook in transport layer doesn't really make the fw assert. Moving this hook to the op_mode allows to use the fw API to actually send a host command that will make the fw assert. Change the restart_fw module parameter to be a boolean on the way. Signed-off-by:
-
Emmanuel Grumbach authored
The firmware tells the driver to what MACs the received frame belongs (based on the time slot in which it was received). Note that there can be several MACs if they share the same binding. Signed-off-by:
-
Emmanuel Grumbach authored
This will allow to track how BT core updates the driver. This is required to debug the BT Coexistence mechanism. Signed-off-by:
-
Emmanuel Grumbach authored
When BT traffic load gets higher, we want to avoid using the shared antenna. In order to do so, we need to tell the AP that we don't support MIMO any more, or at least not all the time: in short, use the SMPS to achieve this. Signed-off-by:
-
Emmanuel Grumbach authored
The BT-Coex notification is sent by the fw when there are updates wrt. BT activity. Driver action might be taken based on the info in this notification. For now, update the Ack/Cts_kill_msk if HID / SCO / A2DP profiles are active. Signed-off-by:
-
Emmanuel Grumbach authored
Send the PRIO table before the calibrations. This table tells the fw what priority to give to what (WiFi / BT) according to events. Send a hardcoded BT_COEX command to the fw to enable basic BT coexistence. Signed-off-by:
-
Emmanuel Grumbach authored
This is the API to tell the fw to handle the BT Coexistence. Signed-off-by:
-
Emmanuel Grumbach authored
Then the transport can print it nicely in its debug prints. Signed-off-by:
-
Johannes Berg authored
If all the pieces of iwlwifi are built into the kernel then there's no need for it to export its symbols to other modules, so prevent that. Reviewed-by:
-
Johannes Berg authored
Now that we have two drivers (DVM and MVM) stop selecting the DVM one (but make it default) and allow enabling only the MVM driver if so desired. Add a warning for the case of having neither DVM nor MVM enabled -- that's useless. Reviewed-by:
-
Dor Shaish authored
Signed-off-by:
Dor Shaish <dor.shaish@intel.com> Reviewed-by:
-
Johannes Berg authored
When modifying a MAC, we update its beacon system time which is taken as a base to calculate TBTT. The firmware doesn't use the new timestamp because the time is never used after the MAC and broadcast station were added, but it is safer to not rely on this and avoids the overhead of reading the register every time the MAC is updated. Reviewed-by:
-
Emmanuel Grumbach authored
Newer devices can work on different buses. This means that their configuration can be shared between different buses. Hence the configuration structures should exported to all the buses and not only to PCIE. Change this. Note that this requires all the fields to be the same amongst the buses. If differences will appear, we can always define a part that is bus dependent. Today, this is not needed. Signed-off-by:
-
Emmanuel Grumbach authored
All the data coming from the fw must have a length that is multiple of 4. This doesn't change anything to the way we handle the notification. Signed-off-by:
-
Eytan Lifshitz authored
Update the NVM parsing functions to add VHT capabilities; they are only added for 5 GHz, of course. This assumes that all devices with NVM reading (rather than EEPROM) that support 5 GHz have VHT, which is true right now. Signed-off-by:
Eytan Lifshitz <eytan.lifshitz@intel.com> Reviewed-by:
-
Johannes Berg authored
When a station is added, we need to tell the firmware what the SMPS settings and number of streams are. After having the initial data, the firmware will track future changes by itself. Reviewed-by:
-
Johannes Berg authored
With remote wake, the firmware creates a TCP connection and sends some configurable data on it, until a special TCP data packet from the server is received that triggers a wakeup. The configuration is a bit tricky because it is based on packet pattern matching but this is hidden in the driver and the exposed API in cfg80211 is just based on the required TCP connection parameters. Reviewed-by:
-
Wei Yongjun authored
This removes an open coded simple_open() function and replaces file operations references to the function with simple_open() instead. Signed-off-by:
Wei Yongjun <yongjun_wei@trendmicro.com.cn> Reviewed-by:
-
Johannes Berg authored
Instead of using (char *)__get_dynamic_array use __get_str. The latter is actually a macro that expands to the former in the code, but trace-cmd in userspace can parse __get_str only. Reviewed-by:
-
Johannes Berg authored
All hardware after 4965 supports this. It's likely that it wasn't set because for 4965 it was irrelevant (HT is only supported on 5 GHz there) and then never updated. Reviewed-by:
-
Emmanuel Grumbach authored
mac80211 tells us when we need to dump the frames from the AGG queue instead of releasing them as single MPDUs. Being able to differentiate between the different cases (IEEE80211_AMPDU_TX_STOP_*) allows us to handle races better. When the station is removed, mac80211 asks to flush and removes the station right away. This allows to avoid a case where we still have frames in AGG queues, but the station has been remove already. Note that we can have frames on the shared queues, but this is not a problem: the station in the fw will be kept until all the frames on the shared queues have been drained. AGG queues are a special case since they are dynamically allocated. Signed-off-by:
-
Ilan Peer authored
Use the number of addresses (max 5) from the NVM instead of limiting to 2 artificially. Signed-off-by:
-
Beni Lev authored
There's a bug that causes the rate scaling to get stuck when it has to use single-stream rates with a peer that can do GF and SGI; the two are incompatible so we can't use them together, but that causes the algorithm to not work at all, it always rejects updates. Disable greenfield for now to prevent that problem. The MVM driver currently only works on devices that don't support greenfield anyway, but better be safe and not allow us to forget about this. Signed-off-by:
Beni Lev <beni.lev@intel.com> Reviewed-by:
-
Beni Lev authored
The 7000 series devices don't support HT greenfield mode so don't advertise or use it. Signed-off-by:
-
Ilan Peer authored
1. For P2P Device filter in only probe requests. 2. For station mode filter in all group cast frames, and in addition beacons as long as we are not associated. 3. For AP/GO filter in all group cast and in addition probe requests. Signed-off-by:
-
Dor Shaish authored
The current fw doesn't currently support cts to self. There is a bug in the fw that prevents us from using cts to self. Use full protection (including RTS) for now. Signed-off-by:
-
Emmanuel Grumbach authored
When we stop an AGG session, we need to look at the sequence numbers in in the private area of the ieee80211_sta struct. This allows us to know is the queue is empty. To get access to this private area, we use fw_id_to_mac_id that maps sta_id (index of the STA in fw table) to ieee80211_sta. When the STA exists in fw, but not in mac80211, we set an ERR ptr in fw_id_to_mac_id. But if we first set an ERR ptr to fw_id_to_mac_id, and only then flush the queues, then we won't be able to access the sequence numbers in ieee80211_sta from the reclaim flow. This means that we will never be able to release an AGG queue when a station is deleted. So first, flush the queue. That will let the reclaim flow call iwl_mvm_check_ratid_empty which will disable the AGG queue as needed, and only then, remove the mapping in fw_id_to_mac_id. Signed-off-by:
-
Emmanuel Grumbach authored
We didn't check that we allowed to start Tx AGG. This can possibly be avoided by a module parameter. Fix that. Signed-off-by:
-
Emmanuel Grumbach authored
7000.c was released as GPL only by mistake: it should be dual licensed - GPL / BSD. The file that contains the license in the kernel is COPYING and not LICENSE.GPL. Signed-off-by:
-
Johannes Berg authored
This is needed to resolve some conflicts that would otherwise happen between wireless-next and the code here. Signed-off-by:
-
- 04 Mar, 2013 4 commits
-
-
Emmanuel Grumbach authored
This flow happens when we get a failed single Tx response on an AMPDU queue. In this case, the frame won't be sent any more. So we need to move the window on the recipient side. This is done by a BAR. Now if we are in the following case: 10, 12 and 13 are ACKed and 11 isn't. 10 11 12 13. V X V V Then, 11 will be sent 16 times as an MPDU (as oppsed to A-MPDU). If this failed, we are entering the flow described above. So we need to send a BAR with ssn = 12. But in this case, the scheduler will tell us to free frames up to 13 (included). So, it is perfectly possible to get a failed single Tx response on an AMPDU queue that makes the scheduler's ssn jump by more than 1 single packet. Signed-off-by:
-
Emmanuel Grumbach authored
Make the rssi more accurate by taking in count per-chain AGC values. Without this, the RSSI reports inaccurate values. Signed-off-by:
-
Emmanuel Grumbach authored
Since the device is being restarted, all the Rx / Tx Block Ack sessions are been wiped out by the driver. So ignore the requests from mac80211 that stops Tx agg while reconfiguring the device. Note that stopping a non-existing Rx BA session is harmless, so just honor mac80211's request. Signed-off-by:
-
Dor Shaish authored
This fix removes the override of calibration request values sent to the FW. Due to that, the sending of default values to now implemented calibrations is removed. Signed-off-by:
-
