Helmut Schaa [Thu, 4 Nov 2010 19:36:59 +0000 (20:36 +0100)]
rt2x00: Add TXOP_CTRL_CFG register definition
Remove the magic value initialisation of the TXOP_CTRL_CFG register by
defining its fields and using them during intialisation. The field
RESERVED_TRUN_EN is referred to as reserved, however it is set to 1 by
the legacy drivers. Hence, do the same.
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com> Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Sun, 7 Nov 2010 13:59:39 +0000 (14:59 +0100)]
ath9k: rework tx queue selection and fix queue stopping/waking
The current ath9k tx queue handling code showed a few issues that could
lead to locking issues, tx stalls due to stopped queues, and maybe even
DMA issues.
The main source of these issues is that in some places the queue is
selected via skb queue mapping in places where this mapping may no
longer be valid. One such place is when data frames are transmitted via
the CAB queue (for powersave buffered frames). This is made even worse
by a lookup WMM AC values from the assigned tx queue (which is
undefined for the CAB queue).
This messed up the pending frame counting, which in turn caused issues
with queues getting stopped, but not woken again.
To fix these issues, this patch removes an unnecessary abstraction
separating a driver internal queue number from the skb queue number
(not to be confused with the hardware queue number).
It seems that this abstraction may have been necessary because of tx
queue preinitialization from the initvals. This patch avoids breakage
here by pushing the software <-> hardware queue mapping to the function
that assigns the tx queues and redefining the WMM AC definitions to
match the numbers used by mac80211 (also affects ath9k_htc).
To ensure consistency wrt. pending frame count tracking, these counters
are moved to the ath_txq struct, updated with the txq lock held, but
only where the tx queue selected by the skb queue map actually matches
the tx queue used by the driver for the frame.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Reported-by: Björn Smedman <bjorn.smedman@venatech.se> Signed-off-by: John W. Linville <linville@tuxdriver.com>
The Mandriva patch seems to stem from 2.6.14, so much for their
upstreaming effort.
Didn't find another Linux reference of it, just an omnious
"USB\VID_1044&PID_8004" from GigabyteZD1201U.INF for
Gigabyte GN-WLBZ101 802.11b USB Adapter, which matches the
Mandriva patch comment.
Aboves file also lists an "USB\VID_1044&PID_8006", which I have
kept appart as this "Gigabyte GN-WBZB-M 802.11b USB Adapter"
didn't show up in googling.
Signed-off-by: maximilian attems <max@stro.at> Signed-off-by: John W. Linville <linville@tuxdriver.com>
"These USB ID came from Palnex <http://www.planex.co.jp/>
Worked fine." says Mandriva patch for their 2.6.32 and earlier.
Web has evidence for both id's to work, so just add them upstream:
http://www.mail-archive.com/zd1211-devs@lists.sourceforge.net/msg00507.html
http://ubuntuforums.org/showthread.php?t=473046
Signed-off-by: Go Taniguchi <go@turbolinux.co.jp> Signed-off-by: maximilian attems <max@stro.at> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rtl8187: restore anaparam registers after reset with 8187B
Current 8187B initialization misses anaparam registers restore after
8187 reset. This causes ANAPARAM register to stay zeroed out (ANAPARAM2
kept its value on my tests). To avoid this, call rtl8187_set_anaparam
right after chip reset (to be on the safe side, as it makes sure we
restore all ANAPARAM registers).
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Cc: seno <senada@t-online.de> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Usually you set RTL818X_CONFIG3_ANAPARAM_WRITE when you are going to
change/write ANAPARAM registers. But in current initialization of
RTL8187B there is a place where ANAPARAM_WRITE bit is set without any
ANAPARAM register being written, without reason, so remove it.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Cc: seno <senada@t-online.de> Signed-off-by: John W. Linville <linville@tuxdriver.com>
There are repeated calls for anaparam on/off sequence in the code.
Consolidate the common code in rtl8187_set_anaparam and use it where
needed.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rtl8187: don't set RTL818X_CONFIG3_GNT_SELECT flag on 8187B
The GNTSel bit should only concern pci devices by looking at RTL8180
spec, which is not the case of 8187B. Also testing shows that trying to
set this bit fails, a subsequent read from the register after trying to
set it shows that the bit isn't set, seems the hardware ignores it,
which makes sense. This setting was a left over from Realtek sources.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rtl8187: move pll reset at start out of ANAPARAM write
On 8187B start, comment about pll reset, and move it out of ANAPARAM
write sequence, so that code is more readable.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rtl8187: avoid redundant write to register FF72 (RFSW_CTRL)
The table with misc register initialization was setting it, and later
on we would set it again with a explicity call to rtl818x_iowrite16_idx.
Remove duplicate initialization from the register table.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rtl8187: fix wrong register initialization in 8187B
We were using wrong address for BRSR (Basic Rate Set Register) while
initializing its value, comparing with Realtek sources, for 8187B case.
Also, the same register is initialized in rtl8187b_reg_table, so remove
the duplicate initialization from the table.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
rtl8187: remove setting of beacon/atim registers from initialization
On 8187B path, we set a initial value for beacon interval and atim
window on initialization. But this isn't needed, since same setup is
done on rtl8187_config.
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This removes redundant write to Auto Rate Fallback Register on RTL8187B.
The same value was being written twice in the same function. Avoid this
removing the duplicate initialization on rtl8187b_reg_table, and also
add comment for this write (information from Realtek source).
Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Acked-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
AP firmware uses xmitcontrol to differentiate between AMPDU
and non-AMPDU frames. As the support for AMPDU is not yet
added, set xmitcontrol to non-AMPDU for all tx frames for AP
firmware. This field will be set to indicate ampdu/non-ampdu
frames when tx AMPDU support is added.
Signed-off-by: Pradeep Nemavat <pnemavat@marvell.com> Signed-off-by: Brian Cavagnolo <brian@cozybit.com> Acked-by: Lennert Buytenhek <buytenh@wantstofly.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch fixes a possible lengthy stall if the device
is operating as an experimental 11n AP and an STA
[during heavy txrx action] suddenly signalized to go
off-channel (old NetworkManager), or (sleep - which is
unlikely, because then it wouldn't be *active* at all!?).
Because the driver has to manage the BA Window, the
sudden PSM transition can leave active uplink BA
sessions to the STA in a bad state and a proper
cleanup is needed.
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
carl9170: fix spurious restart due to high latency
RX Stress tests of unidirectional bulk traffic with
bitrates of up to 220Mbit/s have revealed that the
fatal-event recovery logic [which was solely triggered
by an out-of-rx-buffer situation] is too aggressive.
The new method now "pings" the device and then
decides - based on the response - whenever
a restart is needed or not.
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch imports all shared header changes
from carl9170fw.git.
* add some strategic __aligned(4).
This allows the compiler generate optimized code for
architectures which can't access (unaligned/packed)
data efficiently.
("ath9k_hw: optimize all descriptor access functions")
* add a forgotten __CARL9170FW__ ifdef around
a private firmware-internal struct.
* GET_VAL macro helper
Very useful for extracting data out of the
bit-packed PHY registers.
* cosmetic changes
e.g.: _CCA_MINCCA_ to just _CCA_MIN_.
* version bump 1.8.8.3 -> 1.9.0.
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Blaise Gassend [Thu, 28 Oct 2010 09:01:24 +0000 (02:01 -0700)]
mac80211_hwsim: Incorporate txpower into rssi
Up to now mac80211_hwsim has been reporting an rssi of -50. This patch
improves the model slightly by returning txpower-50. This makes it
easy to stimulate tests that need to see a varying rssi.
Signed-off-by: Blaise Gassend <blaise@willowgarage.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Clearing the per packet TX AGC for the RTL8187B device appears to
increase its overall TX power. This allows the device to associate and a
connection to be established using APs a little further away.
This is in accordance to what is done for RTL8187L devices and also what
Realtek drivers do.
Tested-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com> Cc: linux-wireless@vger.kernel.org Cc: Larry Finger <Larry.Finger@lwfinger.net> Cc: Rogerio Luz Coelho <rogluz.news@gmail.com> Cc: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Cc: Hin-Tak Leung <hintak.leung@gmail.com> Cc: seno <senada@t-online.de> Tested-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Juuso Oikarinen [Tue, 26 Oct 2010 11:24:39 +0000 (13:24 +0200)]
wl1271: Check interface state in op_* functions
Check the state of the interface on op_* function so we don't try to access
the hardware in when its off.
The mac80211 may call these in some corner cases related, for instance, to
the hardware recovery procedure. These accesses cause a kernel crash on at
least some SDIO devices, because the bus is not properly claimed in that
scenario.
Juuso Oikarinen [Tue, 26 Oct 2010 11:24:38 +0000 (13:24 +0200)]
wl1271: Fix scan failure detection
In scan_complete_work, because the mutex is released before accessing the
scan->failed flag, it is possible for unfounded hardware recovery rounds
to be executed.
Juuso Oikarinen [Mon, 25 Oct 2010 09:24:29 +0000 (11:24 +0200)]
wl1271: Fix RX path stall
The wl1271_rx function loops through packets in an aggregated buffer. Each
packet in the buffer is handled by a call to wl1271_rx_handle_data, which will
fail if skb memory allocation fails or production mode is enabled. These
failures currently prevent the rx counters to be incremented, thus causing the
rx loop to run forever.
Fix this by ignoring error codes reported wl1271_rx_handle_data function.
This essentially means that frames will be dropped in production mode, which
is the intetion, and frames will be dropped if memory allocation fails, which
is a decent way to recover from that situation.
Luciano Coelho [Wed, 20 Oct 2010 12:15:52 +0000 (15:15 +0300)]
wl1271: exit ELP mode when setting enabled rates in tx
This bug was being triggered by a call to acx_rate_policies in tx_work
without calling ps_elp_wakeup first. If we have full PSM enabled, this
happens rather often, immediately after association.
Reported-by: Tuomas Katila <ext-tuomas.2.katila@nokia.com> Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com> Tested-by: Tuomas Katila <ext-tuomas.2.katila@nokia.com>
Teemu Paasikivi [Thu, 14 Oct 2010 09:00:04 +0000 (11:00 +0200)]
wl1271: Fix warning about unsupported RX rate
While scanning, it is possible that beacon and probe response frames are
received on other band than configured to the driver. In rx status
handling this has caused "Unsupported RX rate from HW" warnings. This
patch changes the wl1271_rate_to_index function to take the band of the
received frame as a parameter instead of using value configuret to
wl->band.
Shahar Levi [Wed, 13 Oct 2010 14:09:41 +0000 (16:09 +0200)]
wl1271: 11n Support, functionality and configuration ability
Add 11n ability in scan, connection and using MCS rates.
The configuration is temporary due to the code incomplete and
still in testing process. That plans to be remove in the future.
Signed-off-by: Shahar Levi <shahar_levi@ti.com> Reviewed-by: Luciano Coelho <luciano.coelho@nokia.com> Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
Ido Yariv [Tue, 12 Oct 2010 12:49:12 +0000 (14:49 +0200)]
wl1271: Fix TX queue low watermark handling
The number of entries in the TX queue is compared to the low watermark
value each time TX completion interrupts are handled.
However, the fact that a TX completion arrived does not necessarily mean
there are any less skbs in the TX queue.
In addition, a TX completion interrupt does not necessarily mean that there
are any new available TX blocks. Thus, queuing TX work when the low
watermark is reached might not be needed.
Fix this by moving the low watermark handling to the TX work function,
and avoid queuing TX work in this case.
Ido Yariv [Tue, 12 Oct 2010 12:49:10 +0000 (14:49 +0200)]
wl1271: Fix TX starvation
While wl1271_irq_work handles RX directly (by calling wl1271_rx), a different
work is scheduled for transmitting packets. The IRQ work might handle more than
one interrupt during a single call, including multiple TX completion
interrupts. This might starve TX, since no packets are transmitted until all
interrupts are handled.
Fix this by calling the TX work function directly, instead of deferring
it.
Ido Yariv [Tue, 12 Oct 2010 12:49:09 +0000 (14:49 +0200)]
wl1271: TX aggregation optimization
In case the aggregation buffer is too small to hold all available packets,
the buffer is transferred to the FW and no more packets are aggregated.
Although there may be enough available TX blocks, no additional packets will
be handled by the current TX work.
Fix this by flushing the aggregation buffer when it's full, and continue
transferring packets as long as there are enough available TX blocks.
Wey-Yi Guy [Sat, 23 Oct 2010 16:15:44 +0000 (09:15 -0700)]
iwlagn: turn dynamic smps on while BT is on
While BT is on and doing iscan and/or pscan, BT is in listen
mode which will impact WiFi throughput, we need to
enable dynamic smps in order to improve the rx throughput.
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Sat, 23 Oct 2010 16:15:43 +0000 (09:15 -0700)]
iwlwifi: implement switching iftype while up
Implement switching the interface while an
interface is up in iwlwifi. Interfaces have
to stay on the context they were created on.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Sat, 23 Oct 2010 16:15:42 +0000 (09:15 -0700)]
iwlwifi: fix IBSS beaconing
My previous patch to clean up all RXON handling
inadvertently broke IBSS because it failed to
take into account that unlike in AP mode, IBSS
requires beacons to be sent only after setting
the RXON assoc.
Fix this, clean up the code a bit, improve the
error checking around this, and also react to
beacon changes in IBSS mode from mac80211.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Sat, 23 Oct 2010 16:15:41 +0000 (09:15 -0700)]
iwlagn: new RXON processing for modern devices
In order to simplify the flow, and make new
enhancements easier, separate out the RXON
processing for modern AGN (5000 and newer)
from RXON processing for the older 3945 and
4965 devices. Avoid changing these old ones
to avoid regressions and move their code to
a new file (iwl-legacy.c). 4965 gets the
commit_rxon that used to be common for all
AGN devices, but with removed PAN support.
The new RXON processing is more central and
does more work in committing, so that it is
easier to follow.
To make it more evident what is split out
for legacy, split the necessary operations
for that into a new struct iwl_legacy_ops.
Those parts that still exist in the new AGN
code don't need to be parametrized.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg [Sat, 23 Oct 2010 16:15:39 +0000 (09:15 -0700)]
iwlwifi: make mac80211 ops a device config
In the future, 4965 and modern AGN devices will
need to have different mac80211 callbacks since
they have different capabilities. Prepare for
that by making the mac80211 operations a device
config.
Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath9k: content DMA start / stop through the PCU lock
This helps align resets / RX enable & disable / TX stop / start.
Locking around the PCU is important to ensure the hardware doesn't
get stale data when working with DMA'able data.
This is part of a series of patches which fix stopping
TX DMA completley when requested on the driver.
For more details about this issue refer to this thread:
Tested-by: Ben Greear <greearb@candelatech.com> Cc: Kyungwan Nam <kyungwan.nam@atheros.com> Cc: stable@kernel.org Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
The PCU lock should be used to contend TX DMA as well,
this will be done next.
This is part of a series of patches which fix stopping
TX DMA completley when requested on the driver.
For more details about this issue refer to this thread:
Tested-by: Ben Greear <greearb@candelatech.com> Cc: Kyungwan Nam <kyungwan.nam@atheros.com> Cc: stable@kernel.org Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
The new PCU lock is better placed so we can just contend
against that when trying to reset hardware.
This is part of a series of patches which fix stopping
TX DMA completley when requested on the driver.
For more details about this issue refer to this thread:
Tested-by: Ben Greear <greearb@candelatech.com> Cc: Kyungwan Nam <kyungwan.nam@atheros.com> Cc: stable@kernel.org Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Hauke Mehrtens [Sat, 23 Oct 2010 17:51:32 +0000 (19:51 +0200)]
carl9170: fix typos
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de> Acked-by: Christian Lamparter <chunkeey@googlemail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
iwlwifi: avoid commit rxon during scan in iwl_set_no_assoc
Currently we are canceling scan when changing BSSID. Behave the same
when changing association and beacon enablement, to avoid committing
rxon during scan in iwl_set_no_assoc().
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Do not set power mode when scanning, and defer that when scan finish.
We still set power mode in force case i.e. when device is overheated.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
iwlwifi: avoid commit rxon during scan in iwlagn_bt_traffic_change_work
Avoid sending commands to firmware (including commit_rxon) when scan
is pending and we are calling iwlagn_bt_traffic_change_work simultaneously.
Also comment some innocent race conditions.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
iwlwifi: avoid commit rxon during scan in iwlagn_configure_filter
Almost anywhere in the code we avoid committing rxon while performing
scan, and make rxon commit when scan complete. However in some places
in the code we do not follow that rule. This patch fix that problem in
iwlagn_configure_filter().
Since we do not commit directly in iwl3945_configure_filter, we can
also do the same for agn, so I just remove iwlcore_commit_rxon()
function and add a comment. Also change comment for iwl3945.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
According to comment in iwl_bg_scan_completed, setting tx power should
be deferred during pending scan, but we are not doing this.
This patch change code to really defer setting tx power after scan
complete. Additionally refactor iwl_set_tx_power code and call
lib->send_tx_power() directly from iwlagn_commit_rxon.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
On 5xxx and 6xxx change to send tx_power_cmd command synchronously,
to do not start other commands when setting tx power is pending.
We currently do the same for 4956 and 3945.
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
iwlwifi: warn when send tx power settings during scan
Add WARN_ONCE when scanning is pending. Use STATUS_SCAN_HW bit since we
can have scan canceled or completed but STATUS_SCANNING bit still set.
v1 -> v2: replace EIO to EAGAIN
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com> Acked-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Thu, 21 Oct 2010 00:47:25 +0000 (02:47 +0200)]
mac80211: minstrel_ht - reduce the overhead of rate sampling
- reduce the number of retransmission attempts for sample rates
- sample lower rates less often
- do not use RTS/CTS for sampling frames
- increase the time between sampling attempts
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
We have seen several DMA races when we race against
stopping and starting the PCU. I suspect that when
we cannot stop the PCU we may hit some of these same
races so warn against them for now but only when
debugging (CONFIG_ATH_DEBUG) is enabled.
If you run into this warning and are a developer,
please fix the cause of the warning. The potential
here, although I cannot prove yet, is that the DMA
engine can be confused and start writing to a buffer
that was already DMA'd before and at least the kernel
assumes is not being accessed by hardware anymore.
Cc: Ben Greear <greearb@candelatech.com> Cc: Kyungwan Nam <kyungwan.nam@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
To be used to throw out warnings only for developers.
This can be used by some corner cases that developers
already know can be hit but developers want to address
so to avoid spewing out a warning this can only be
enabled with CONFIG_ATH_DEBUG enabled.
Cc: Ben Greear <greearb@candelatech.com> Cc: Kyungwan Nam <kyungwan.nam@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Cc: Easwar Krishnan <easwar.krishnan@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
cfg80211: add debug print when processing a channel
In the worst case you are seeing really odd things you want
more information than what is provided right now, for those
that insist and want debug info through CONFIG_CFG80211_REG_DEBUG
provide a print of when we are processing a channel and with what
regulatory rule.
Cc: Easwar Krishnan <easwar.krishnan@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
cfg80211: add debug print when disabling a channel on a custom regd
Cc: Easwar Krishnan <easwar.krishnan@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
cfg80211: add debug prints for when we ignore regulatory hints
This can help with debugging issues. You will only see
these with CONFIG_CFG80211_REG_DEBUG enabled.
Cc: Easwar Krishnan <easwar.krishnan@atheros.com> Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
After a module loads you will have loaded the world roaming regulatory
domain or a custom regulatory domain. Further regulatory hints are
welcomed and should be respected unless the regulatory hint is coming
from a country IE as the IEEE spec allows for a country IE to be a subset
of what is allowed by the local regulatory agencies.
So disable all channels that do not fit a regulatory domain sent
from a unless the hint is from a country IE and the country IE had
no information about the band we are currently processing.
This fixes a few regulatory issues, for example for drivers that depend
on CRDA and had no 5 GHz freqencies allowed were not properly disabling
5 GHz at all, furthermore it also allows users to restrict devices
further as was intended.
If you recieve a country IE upon association we will also disable the
channels that are not allowed if the country IE had at least one
channel on the respective band we are procesing.
This was the original intention behind this design but it was
completely overlooked...
Cc: David Quan <david.quan@atheros.com> Cc: Jouni Malinen <jouni.malinen@atheros.com>
cc: Easwar Krishnan <easwar.krishnan@atheros.com> Cc: stable@kernel.org Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
cfg80211: fix allowing country IEs for WIPHY_FLAG_STRICT_REGULATORY
We should be enabling country IE hints for WIPHY_FLAG_STRICT_REGULATORY
even if we haven't yet recieved regulatory domain hint for the driver
if it needed one. Without this Country IEs are not passed on to drivers
that have set WIPHY_FLAG_STRICT_REGULATORY, today this is just all
Atheros chipset drivers: ath5k, ath9k, ar9170, carl9170.
This was part of the original design, however it was completely
overlooked...
Cc: Easwar Krishnan <easwar.krishnan@atheros.com> Cc: stable@kernel.org Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Larry Finger [Wed, 20 Oct 2010 14:59:33 +0000 (09:59 -0500)]
ssb: Clear RETRY_TIMEOUT in PCI Configuration for normal devices
MMIO log traces obtained using the Broadcom wl hybrid driver show that
the RETRY_TIMEOUT register (0x41) in PCI configuration space is cleared
if non-zero. Similar code found in other drivers such as ipw2100 show
this operation is needed to keep PCI Tx retries from interfering with
C3 CPU state. There are no known cases where omission of this code has
caused a problem, but this patch is offered just in case such a situation
occurs.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath5k: make ath5k_update_bssid_mask_and_opmode() static
This fixes this sparse warning:
CHECK drivers/net/wireless/ath/ath5k/base.c
drivers/net/wireless/ath/ath5k/base.c:569:6: warning: symbol
'ath5k_update_bssid_mask_and_opmode' was not declared. Should it be static?
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
CHECK drivers/net/wireless/ath/key.c
drivers/net/wireless/ath/key.c:110:6: warning: symbol
'ath_hw_set_keycache_entry' was not declared. Should it be static?
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Wed, 20 Oct 2010 01:08:53 +0000 (03:08 +0200)]
ath9k_hw: extend ath9k_hw_set_txpowerlimit to test channel txpower
ath9k_hw_set_txpowerlimit gets an extra boolean parameter that - if set -
causes the rate txpower table and the regulatory limit to be calculated
and stored, without changing hardware registers.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Fri, 15 Oct 2010 23:01:48 +0000 (01:01 +0200)]
ath9k_hw: optimize all descriptor access functions
Because all of the descriptor data structures are marked as __packed, GCC
assumes the worst case wrt. alignment and generates unaligned load/store
instructions on MIPS for access to all fields.
Since descriptors always have to be 4-byte-aligned, we can just mark the
data structures with __aligned(4), which allows GCC to generate much more
efficient code.
Verified through disassembly and OProfile comparisons.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
The ath9k driver uses the legacy PCI power management (suspend
and resume) callbacks that apparently cause intermittent problems
to happen (the adapter sometimes doesn't resume correctly on my
Acer Ferrari One). Make it use the new PCI PM and let the PCI core
code handle the PCI-specific details of power transitions.
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> Signed-off-by: John W. Linville <linville@tuxdriver.com>
CHECK drivers/net/wireless/ath/ath9k/debug.c
drivers/net/wireless/ath/ath9k/debug.c:548:34: warning: symbol 'aphy' shadows an earlier one
drivers/net/wireless/ath/ath9k/debug.c:491:26: originally declared here
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Fri, 15 Oct 2010 18:03:32 +0000 (20:03 +0200)]
ath9k: remove a redundant call to ath9k_hw_gettsf32
When the timer_next argument to ath9k_gen_timer_start is behind the tsf value,
tsf + timer_period is used, which is what ath_btcoex_period_timer was setting
it to.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Fri, 15 Oct 2010 18:03:31 +0000 (20:03 +0200)]
ath9k: optimize/fix ANI RSSI processing
ANI needs the RSSI average only in station mode, and only for tracking
the signal strength of beacons of the AP that it is connected to.
Adjust the code to track on the beacon RSSI, and store the average of that
in the ath_wiphy struct.
With these changes, we can get rid of this extra station lookup in the
rx path, which saves precious CPU cycles.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Fri, 15 Oct 2010 18:03:30 +0000 (20:03 +0200)]
ath9k_hw: optimize tx status descriptor processing
Disassembly shows, that at least on MIPS, the compiler generates a lot of
memory accesses to the same location in the descriptor field parsing.
Since it is operating on uncached memory, this can be quite expensive in
this hot path.
Change the code a bit to help the compiler optimize it properly, and get
rid of some unused fields in the ath_tx_status struct.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Fri, 15 Oct 2010 18:03:29 +0000 (20:03 +0200)]
ath9k_hw: small optimization in ar9002_hw_get_isr
ah->config.rx_intr_mitigation does not need to be checked before checking
the rx interrupt mask for AR_ISR_RXMINTR or AR_ISR_RXINTM, as those
interrupts will be masked out if rx interrupt mitigation is disabled.
Avoid reading AR_ISR_S5_S twice by reordering the code to be more concise.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
Felix Fietkau [Mon, 8 Nov 2010 19:54:47 +0000 (20:54 +0100)]
ath9k_hw: optimize interrupt mask changes
OProfile showed that ath9k was spending way too much time in
ath9k_hw_set_interrupts. Since most of the interrupt mask changes only
need to globally enable/disable interrupts, it makes sense to split
this part into separate functions, replacing all calls to
ath9k_hw_set_interrupts(ah, 0) with ath9k_hw_disable_interrupts(ah).
ath9k_hw_set_interrupts(ah, ah->imask) only gets changed to
ath9k_hw_enable_interrupts(ah), whenever ah->imask was not changed
since the point where interrupts were disabled.
Signed-off-by: Felix Fietkau <nbd@openwrt.org> Signed-off-by: John W. Linville <linville@tuxdriver.com>
projects / karo-tx-linux.git / log