- 27 Oct, 2009 40 commits
-
-
Juuso Oikarinen authored
Add a configuration structure for TX path parameters, and set defalt configuration values there. Signed-off-by:
Juuso Oikarinen <juuso.oikarinen@nokia.com> Reviewed-by:
Luciano Coelho <luciano.coelho@nokia.com> Signed-off-by:
John W. Linville <linville@tuxdriver.com>
-
Juuso Oikarinen authored
Add a configuration structure for RX path parameters, and set default configuration values there. Signed-off-by:
-
Juuso Oikarinen authored
In order to make the firmware configuration more manageable, collect hardcoded configuration values into one data structure, and set default values there. Add the SoftGemini BT/WLAN coex paramters into the config structure. Signed-off-by:
-
Juuso Oikarinen authored
If the 38.4MHz reference clock is configured to the firmware, it crashes on boot. Configuring an experimental 38.4MHz in XTAL mode allows the firmware to boot, and everything appears to work. Signed-off-by:
-
Zhu Yi authored
This switches the iwlwifi driver to use paged skb from linear skb for Rx buffer. So that it relieves some Rx buffer allocation pressure for the memory subsystem. Currently iwlwifi (4K for 3945) requests 8K bytes for Rx buffer. Due to the trailing skb_shared_info in the skb->data, alloc_skb() will do the next order allocation, which is 16K bytes. This is suboptimal and more likely to fail when the system is under memory usage pressure. Switching to paged Rx skb lets us allocate the RXB directly by alloc_pages(), so that only order 1 allocation is required. It also adjusts the area spin_lock (with IRQ disabled) protected in the tasklet because tasklet guarentees to run only on one CPU and the new unprotected code can be preempted by the IRQ handler. This saves us from spawning another workqueue to make skb_linearize/__pskb_pull_tail happy (which cannot be called in hard irq context). Finally, mac80211 doesn't support paged Rx yet. So we linearize the skb for all the management frames and software decryption or defragmentation required data frames before handed to mac80211. For all the other frames, we __pskb_pull_tail 64 bytes in the linear area of the skb for mac80211 to handle them properly. Signed-off-by:
Zhu Yi <yi.zhu@intel.com> Signed-off-by:
-
Luciano Coelho authored
We were using the join command to change some settings when the stack asked us to do it. In many cases they were not needed (and could cause potential problems), so they were removed. In other cases there are ACX commands that can be used instead of using join to reconfigure. Signed-off-by:
-
Luciano Coelho authored
We shouldn't use a join command to change the filter settings while associated. The right way to do it is to use ACX_RX_CFG. Signed-off-by:
-
Juuso Oikarinen authored
Update the magic values of the wl1271 boot time config according to the new reference driver. Signed-off-by:
-
Juuso Oikarinen authored
With the change to the new firmware, there was a change to firmware memory partitioning. Along with that change, the translation of all partitions was unified, and separate functions for reg and mem access became unnecessary. Cleanup the unnecessary functions. Signed-off-by:
-
Juuso Oikarinen authored
Update the TX path block calculation algorithm based on TI reference. Signed-off-by:
-
Juuso Oikarinen authored
Enable and configure smart reflex. This should have a reducing impact on power consumption. Signed-off-by:
-
Juuso Oikarinen authored
The new firmware has fixed a firmware crash problem related to ELP entry. Enable ELP to conserve power. Signed-off-by:
-
Juuso Oikarinen authored
Remove separate interrupt register reading from the interrupt handling routine. This will slightly improve interrupt performance. Signed-off-by:
-
Juuso Oikarinen authored
Updated RefClk configuration based on reference sources. Apparently this change will improve RF performance. Signed-off-by:
-
Juuso Oikarinen authored
Add top register access function. Signed-off-by:
-
Juuso Oikarinen authored
Remove RX workaround which is not needed with newer firmware revisions. This will reduce one SPI register transaction per RX packet. Signed-off-by:
-
Juuso Oikarinen authored
Update the memory regions and memory mapping to support firmware revision 6.1.0.0.241. Signed-off-by:
-
Luciano Coelho authored
We need to listen to HW_AVAILABLE interrupts when using ELP with firmware revision 6.1.0.0.241. Add WL1271_ACX_INTR_HW_AVAILABLE to the interrupts that are enabled by default. Signed-off-by:
-
Luciano Coelho authored
We still don't have proper calibration for our devices, so we are using way too much power for TX, which causes a lot of distortion. This hack hardcodes the txpower to 7dBm. Signed-off-by:
-
Luciano Coelho authored
We don't get any indication from the stack when we have disassociated. In wl1271, it is important to send a CMD_DISCONNECT before joining again, because the firmware cleans some things up. So we check if we're already joined and disconnect if that's the case. Signed-off-by:
-
Luciano Coelho authored
This patch implements the CMD_DISCONNECT command, which should be sent to the firmware when we are disassociated. Signed-off-by:
-
Wey-Yi Guy authored
Channel switch host command has different data structure for different devices. Adding additional structures to support 5000 and 6000 NICs. unlike 4965, starting with 5000 series, the tx power is managed by uCode, there is no tx power db information need to be passing from driver to uCode; but the space needs to be reserved to match uCode expection. 1000 NIC do not support channel switch operation since it is 'bgn' device, there is no need to have data structure defined for it. Signed-off-by:
Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by:
Reinette Chatre <reinette.chatre@intel.com> Signed-off-by:
-
Abhijeet Kolekar authored
Rearrange the code and groups setting of retry_limit and data_retry_limits code together. Make 'data_retry_limit' setting similar to iwlwifi for better readability. Signed-off-by:
Abhijeet Kolekar <abhijeet.kolekar@intel.com> Signed-off-by:
-
Abhijeet Kolekar authored
Remove the ununsed variable data_retry_limit from priv. Signed-off-by:
-
Abhijeet Kolekar authored
Rename iwl3945_tx_cmd variable tx to tx_cmd for better readability. Signed-off-by:
-
Abhijeet Kolekar authored
3945 and 4965 share the functionality for setting RTS and CTS to the tx_cmd. Unify these functions and move the common functionality to core. Signed-off-by:
-
Wey-Yi Guy authored
Adding accumulative statistics counters in iwlwifi driver. Statistics counters are reported by uCode every beacon interval; but can be reset by uCode when needed. The accumulative statistics counters is maintained by driver to keep track of the history of all the counters. Update the ucode stats files in debugfs to display both latest and accumulative counters. Signed-off-by:
-
Wey-Yi Guy authored
Instead of always allocate the max number of tx queue structure, use dynamic allocation based on the number of queues in device configuration. With these changes, device does not have to allocate more memory than the h/w can support. Signed-off-by:
-
Ben Cahill authored
iwl_power_initialize() initializes driver data (not device hardware), and does not need to execute more than once (when the driver initializes). Therefore, it does not belong in iwl3945_apm_init(), which initializes hardware, and may run more than once. Move it to iwl3945_pci_probe(), where it will run only once. This agrees with similar placement in iwl-agn.c's iwl_pci_probe(), although placement under "services" seemed more appropriate than under "mac80211". Signed-off-by:
Ben Cahill <ben.m.cahill@intel.com> Signed-off-by:
-
Ben Cahill authored
iwl3945_rfkill_poll() has been silently calling wiphy_rfkill_set_hw_state() every 2 seconds, regardless of whether hardware RF KILL switch changed state. Call wiphy_rfkill_set_hw_state() only when RFKILL switch changes. Add IWL_DEBUG_RF_KILL log message and documentation. Signed-off-by:
-
Reinette Chatre authored
The sleep_level_override debugfs file is used by the user to request a static power index instead of the dynamic sleep values. Users are expected to provide value from 1 to 5 as an index or -1 to disable it. The problem at the moment is that users can also provide 0 to this file which, together with the value 1, is translated to index 1. This is confusing and even more so when users write 0 to sleep_level_override and then read 1 from it afterwards. Modify checking to treat 0 as invalid. Signed-off-by:
-
Wey-Yi Guy authored
Add more items to sensitivity range table to avoid using hardcoded values. Initialize the table per device since unique per device information is required to perform sensitivity calibration. additional items in sensitivity range table: .barker_corr_th_min: Barker correlation threshold minimum .barker_corr_th_min_mrc: Barker correlation threshold minimum for MRC .nrg_th_cca: Energy threshold for Clear Channel Assessment Barker codes are a technique used in WLAN encoding for transmission. MRC is "Maximal Ratio Combining", a technique for optimally combining the signals from 2 or more receivers to achieve a better signal. Signed-off-by:
-
Wey-Yi Guy authored
"IWL_CMD_QUEUE_NUM" is being defined in multiple places and used by all the devices. move it to iwl-dev.h file and shared by all the devices. Remove "IWL_CMD_FIFO_NUM", replaced by "IWL49_CMD_FIFO_NUM" and IWL50_CMD_FIFO_NUM" Signed-off-by:
-
Ben Cahill authored
Clean up device-specific apm_reset() functions and library infrastructure, now that these reset() functions are no longer being used. Signed-off-by:
-
Ben Cahill authored
Now that we're unconditionally using apm_ops.stop() to reset and power-down the device in __iwl3945_down(), the apm_ops.reset() is redundant. Removing this call will also allow us to remove iwl3945_apm_reset(). Remove unneeded iwl_clear_bit(CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ) because this bit will be set again very soon in iwl3945_hw_txq_ctx_stop() and other following calls. Signed-off-by:
-
Ben Cahill authored
In some cases (e.g. when mac80211 calls iwl_mac_stop() for suspend or user no longer wants device active), device has not been going into low power state via __iwl_down(). apm_ops.reset() does not put device into low power state; instead it resets the device, then puts it into a powered-up state ready to be re-loaded with uCode and re-started. This has needlessly warmed up user's laptops and drained batteries. With current architecture in which mac80211 controls device up/down (including resetting device after firmware errors), there is no need for apm_ops.reset() any more; apm_ops.reset() is basically a combination of apm_ops.stop() and apm_ops.init(). Instead, __iwl_down() now unconditionally places the device into a low-power state via apm_ops.stop(). Device may be re-started via __iwl_up() calling apm_ops.init() as soon as it may be needed (e.g. quickly for firmware errors), but in the meantime, device will stop wasting energy. Note that, even in this low power state, if driver re-enables interrupts, the device retains the ability to sense the hardware RF-KILL switch, and (except for 3945) interrupt the host when it changes. Signed-off-by:
-
Ben Cahill authored
Update iwl3945_apm_init() to set up device registers in sequence most recently recommended by factory. Add resets for APMG interrupts and radio chip, formerly done only in iwl3945_apm_reset(); moving them here assures that apm_init() will do a complete job of preparing hardware not only after platform boot, but also after apm_stop() has executed (due to rfkill, ifconfig down, driver unload, etc.). This is in preparation to completely remove apm_reset(). Add some comments. Signed-off-by:
-
matthieu castet authored
Some airo card don't support power Management [1]. Don't abort suspend with those cards. 00:06.0 Network controller: AIRONET Wireless Communications PC4800 (rev 01) Flags: medium devsel, IRQ 17 Memory at dffffe00 (32-bit, non-prefetchable) [size=128] I/O ports at d000 [size=128] I/O ports at cc00 [size=64] Kernel driver in use: airo Signed-off-by:Matthieu CASTET <castet.matthieu@free.fr> Signed-off-by:
-
Michael Buesch authored
Remove me Signed-off-by:
Michael Buesch <mb@bu3sch.de> Signed-off-by:
-
Michael Buesch authored
This slightly shrinks the structure. Signed-off-by:
-
