openwrt/staging/blogic.git
13 years agort2x00: Fix tx aggregation problems with some clients
Helmut Schaa [Mon, 28 Mar 2011 11:35:21 +0000 (13:35 +0200)]
rt2x00: Fix tx aggregation problems with some clients

Some clients seem to rely upon the reception of BlockAckReqs to flush
their rx reorder buffer. In order to fix aggregation for these clients
rt2x00 should send a BlockAckReq if the transmission of an AMPDU
subframe fails.

Introduce a new flag TXDONE_AMPDU to indicate that this is an AMPDU
subframe and pass IEEE80211_TX_STAT_AMPDU_NO_BACK to mac80211 if an
AMPDU subframe failed during transmission.

This fixes aggregation problems with Intel 5100 Windows STAs (and maybe
others as well).

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>
13 years agort2x00: Add support for the ZyXEL NWD-211AN USB
Ivo van Doorn [Mon, 28 Mar 2011 11:34:50 +0000 (13:34 +0200)]
rt2x00: Add support for the ZyXEL NWD-211AN USB

Add new USB ID

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Fix STBC transmissions to STAs with Rx STBC > 1
Helmut Schaa [Mon, 28 Mar 2011 11:34:15 +0000 (13:34 +0200)]
rt2x00: Fix STBC transmissions to STAs with Rx STBC > 1

For STBC transmissions rt2x00 used the number of RxSTBC streams the
destination STA indicates in its HT capabilities as STBC value in
the TXWI.

However, the legacy drivers and our own comment in rt2800.h suggest
that the STBC field in the TXWI only allows a value of 0 or 1.
The values 2 and 3 are reserved (probably for future devices).

And indeed, STBC transmissions to STAs indicating more then 1
RxSTBC stream fail when the STBC field is set to something >1.

Fix this by only setting the STBC field to 1 when STBC should be
used.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Implement tx power temperature compensation
Helmut Schaa [Mon, 28 Mar 2011 11:33:40 +0000 (13:33 +0200)]
rt2x00: Implement tx power temperature compensation

rt2800 devices should adjust their tx power in accordance with the
eeproms temperature calibration values. Add a new driver callback
gain_calibration that is called every 4 seconds.

The rt2800 gain calibration routine simply runs the tx power
configuration that takes care of calculating the temperature
compensation delta.

We don't need to synchronize the calls to rt2800_config_txpower
as they should all happen from mac80211's single threaded workqueue.

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>
13 years agort2x00: Remove obsolete rt2x00queue_align_payload
Helmut Schaa [Mon, 28 Mar 2011 11:33:04 +0000 (13:33 +0200)]
rt2x00: Remove obsolete rt2x00queue_align_payload

Since commit d1c3a37ceeb1a5ea02991a0476355f1a1d3b3e83 ("mac80211:
clarify alignment docs, fix up alignment") removed the requirement
for a 4-byte aligned payload rt2x00queue_align_payload is obsolete
as mac80211 will align the payload when it passes the frame to the
net stack.

As a result we can remove the call to rt2x00queue_align_payload in the
rx path and since that's the last user we can remove
rt2x00queue_align_payload altogether.

One advantage is that we save some alignment operations for frames
that don't need to be aligned (for example beause they are not passed
to the net stack).

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>
13 years agort2x00: Indention cleanup in rt2800lib
Helmut Schaa [Mon, 28 Mar 2011 11:32:32 +0000 (13:32 +0200)]
rt2x00: Indention cleanup in rt2800lib

Fix the indention in rt2800_compesate_txpower and also fix a typo in
the function name rt2800_compesate_txpower -> rt2800_compensate_txpower.

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>
13 years agort2x00: Don't recalculate HT40 compensation for each rate
Helmut Schaa [Mon, 28 Mar 2011 11:32:01 +0000 (13:32 +0200)]
rt2x00: Don't recalculate HT40 compensation for each rate

Previously the HT40 tx power compensation value was calculated for each
rate. However, the calculation is independent of the tx rate and as such
can be precalculated and just passed in for each rate.

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>
13 years agort2x00: Restructure bw_comp calculation
Helmut Schaa [Mon, 28 Mar 2011 11:31:30 +0000 (13:31 +0200)]
rt2x00: Restructure bw_comp calculation

Move the HT40 check inside the calculation function to make it easier
for a later cleanup.

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>
13 years agort2x00: Remove DRIVER_SUPPORT_WATCHDOG flag
Helmut Schaa [Mon, 28 Mar 2011 11:30:59 +0000 (13:30 +0200)]
rt2x00: Remove DRIVER_SUPPORT_WATCHDOG flag

We can simply check if the driver registered the watchdog callback.
There's no need to have an additional flag for that.

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>
13 years agort2x00: Calculate tx status fifo size instead of hardcoding it
Helmut Schaa [Mon, 28 Mar 2011 11:30:36 +0000 (13:30 +0200)]
rt2x00: Calculate tx status fifo size instead of hardcoding it

Instead of hardcoding the tx status fifo size as 512 calculate it based
on the number of tx queues and the number of entries per queue. Also
round the size up to a power of 2 as kfifo would otherwise round it
down.

On rt2800pci this will increase the kfifo size from 512 bytes to 1024
bytes which is then able to hold the tx status for all entries in all
tx queues.

Furthermore, if the number of tx queues or tx entries changes in the
future (use of the MGMT queue for example) the kfifo size doesn't need
to be updated.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Limit rt2800pci txdone processing to 16 entries at once
Helmut Schaa [Mon, 28 Mar 2011 11:30:09 +0000 (13:30 +0200)]
rt2x00: Limit rt2800pci txdone processing to 16 entries at once

Instead of reporting an unlimited number of tx status reports to
mac80211 stop after 16 frames and reschedule the tx status tasklet.
This allows other tasklets to be run inbetween.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Limit rt2x00pci rxdone processing to 16 entries at once
Helmut Schaa [Mon, 28 Mar 2011 11:29:44 +0000 (13:29 +0200)]
rt2x00: Limit rt2x00pci rxdone processing to 16 entries at once

Instead of receiving an unlimited number of frames, stop after 16
entries and reschedule the rxdone tasklet. This allows other tasklets
to be run inbetween.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agortlwifi: Remove unused/unneeded variables
Larry Finger [Sun, 27 Mar 2011 21:19:57 +0000 (16:19 -0500)]
rtlwifi: Remove unused/unneeded variables

Remove some unused variables and correct spelling errors.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: use IEEE80211_HT_CAP_ macros for 11n cap_info
Marc Yang [Sat, 26 Mar 2011 02:47:02 +0000 (19:47 -0700)]
mwifiex: use IEEE80211_HT_CAP_ macros for 11n cap_info

The hw_dot_11n_dev_cap reported by firmware hw_spec
has different format than the 11n capabilities.
Hence a lot of SET_ and RESET_ bit operation macros
were used to convert the dev_cap format to 11n
capability format. However the locally defined 11n
ht_cap macros are not necessary as we can use
IEEE80211_HT_CAP_ macros directly.

The 32-bit dev_cap bitmap is added as comment to
explain the mapping between firmware and 11n spec.

Some unused macros and unnecessary adapter variables
are also removed.

Signed-off-by: Marc Yang <yangyang@marvell.com>
Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: remove macro SHORT_SLOT_TIME_DISABLED
Bing Zhao [Sat, 26 Mar 2011 02:47:01 +0000 (19:47 -0700)]
mwifiex: remove macro SHORT_SLOT_TIME_DISABLED

and SHORT_SLOT_TIME_ENABLED.
Use WLAN_CAPABILITY_SHORT_SLOT_TIME instead.

Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: remove the dependency on crypto_blkcipher
Felix Fietkau [Fri, 25 Mar 2011 20:39:19 +0000 (21:39 +0100)]
mac80211: remove the dependency on crypto_blkcipher

The only thing that using crypto_blkcipher with ecb does over just using
arc4 directly is wrapping the encrypt/decrypt function into a for loop,
looping over each individual character.
To be able to do this, it pulls in around 40 kb worth of unnecessary
kernel modules (at least on a MIPS embedded device).
Using arc4 directly not only eliminates those dependencies, it also makes
the code smaller.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: remove unnecessary _set_auth functions
Marc Yang [Fri, 25 Mar 2011 03:49:39 +0000 (20:49 -0700)]
mwifiex: remove unnecessary _set_auth functions

mwifiex_set_encrypt_mode()
mwifiex_set_auth_mode()
mwifiex_set_auth()
These functions are confusing and misleading.
And they are really not needed at all.

Some unused definitions are also removed.

Signed-off-by: Marc Yang <yangyang@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: remove helper functions for displaying 11n capabilities
Yogesh Ashok Powar [Fri, 25 Mar 2011 03:49:38 +0000 (20:49 -0700)]
mwifiex: remove helper functions for displaying 11n capabilities

'iw list' is sufficient to retrieve the information which
was displayed by these functions.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: fix suppressing probe responses in ad-hoc mode
Felix Fietkau [Thu, 24 Mar 2011 18:46:18 +0000 (19:46 +0100)]
mac80211: fix suppressing probe responses in ad-hoc mode

The commit "mac80211: reply to directed probes in IBSS" changed ad-hoc
specific code to respond to unicast probe requests, even if
drv_tx_last_beacon returns false, however due to confusion over the
meaning of the IEEE80211_RX_RA_MATCH flag, it also unconditionally
enabled responding to multicast probe requests.
Fix this by explicitly checking for a multicast destination address
instead.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove ath9k_get_channel_edges
Mohammed Shafi Shajakhan [Thu, 24 Mar 2011 13:36:40 +0000 (19:06 +0530)]
ath9k_hw: remove ath9k_get_channel_edges

This function is nowhere used.

Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: remove set11n_virtualmorefrag
Mohammed Shafi Shajakhan [Thu, 24 Mar 2011 10:19:54 +0000 (15:49 +0530)]
ath9k: remove set11n_virtualmorefrag

This does not seems to be used anywhere so remove it.

Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: cleanup few redundant macros
Mohammed Shafi Shajakhan [Thu, 24 Mar 2011 09:06:16 +0000 (14:36 +0530)]
ath9k: cleanup few redundant macros

Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agowifi: Add hwflags to debugfs.
Ben Greear [Wed, 23 Mar 2011 21:04:31 +0000 (14:04 -0700)]
wifi: Add hwflags to debugfs.

Aids debugging wifi behaviour.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlegacy: MAINTAINERS
Stanislaw Gruszka [Fri, 25 Mar 2011 16:59:39 +0000 (17:59 +0100)]
iwlegacy: MAINTAINERS

Add iwlegacy driver to MAINTAINERS.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoMerge branch 'wireless-next-2.6' of git://git.kernel.org/pub/scm/linux/kernel/git...
John W. Linville [Mon, 4 Apr 2011 20:10:36 +0000 (16:10 -0400)]
Merge branch 'wireless-next-2.6' of git://git./linux/kernel/git/iwlwifi/iwlwifi-2.6

13 years agoath9k_hw: remove pCap->tx_triglevel_max
Felix Fietkau [Wed, 23 Mar 2011 19:57:33 +0000 (20:57 +0100)]
ath9k_hw: remove pCap->tx_triglevel_max

It has the same purpose (and value) as ah->config.max_txtrig_level

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove ATH9K_HW_CAP_ENHANCEDPM
Felix Fietkau [Wed, 23 Mar 2011 19:57:32 +0000 (20:57 +0100)]
ath9k_hw: remove ATH9K_HW_CAP_ENHANCEDPM

It is not used anywhere

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove pCap->keycache_size
Felix Fietkau [Wed, 23 Mar 2011 19:57:31 +0000 (20:57 +0100)]
ath9k_hw: remove pCap->keycache_size

Similar to the number of tx queue, the number of keycache entries depends
on the chip and shouldn't be messed with based on EEPROM data.
Remove this field and stick to using AR_KEYTABLE_SIZE

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove pCap->reg_cap
Felix Fietkau [Wed, 23 Mar 2011 19:57:30 +0000 (20:57 +0100)]
ath9k_hw: remove pCap->reg_cap

It is not used anywhere and seems pointless

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove ah->config.ht_enable
Felix Fietkau [Wed, 23 Mar 2011 19:57:29 +0000 (20:57 +0100)]
ath9k_hw: remove ah->config.ht_enable

It is only used in one place, and the device id check that it's based on
can be moved there as well.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove pCap->total_queues
Felix Fietkau [Wed, 23 Mar 2011 19:57:28 +0000 (20:57 +0100)]
ath9k_hw: remove pCap->total_queues

The EEPROM contains a field that can restrict the number of hardware queues,
however this is not only useless (all the known chips contain the same
number of hardware queues), but also potentially dangerous in case of a
misprogrammed EEPROM (could trigger driver crashes), so let's just ignore
it completely.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: turn a few big macros into functions
Felix Fietkau [Wed, 23 Mar 2011 19:57:27 +0000 (20:57 +0100)]
ath9k_hw: turn a few big macros into functions

RF_BANK_SETUP, REG_WRITE_RF_ARRAY and REG_WRITE_ARRAY are way too big,
so they shouldn't be inlined at every single callsite, especially since they
can easily be turned into real functions.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: replace REG_READ+REG_WRITE with REG_RMW
Felix Fietkau [Wed, 23 Mar 2011 19:57:26 +0000 (20:57 +0100)]
ath9k_hw: replace REG_READ+REG_WRITE with REG_RMW

It's easier to read and it slightly decreases code size

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: add a new register op for read-mask-write
Felix Fietkau [Wed, 23 Mar 2011 19:57:25 +0000 (20:57 +0100)]
ath9k_hw: add a new register op for read-mask-write

Reduces the number of calls to register ops. On MIPS this reduces the
ath9k_hw binary size from 321k down to 310k

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: embed the ath_ops callbacks in the ath_hw struct
Felix Fietkau [Wed, 23 Mar 2011 19:57:24 +0000 (20:57 +0100)]
ath9k_hw: embed the ath_ops callbacks in the ath_hw struct

With this change, loading the address to a register read/write function
costs only one pointer dereference instead of two. On MIPS this reduces
ath9k_hw binary size from 326k down to 321k.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211: Add nl80211 event for deletion of a station entry
Jouni Malinen [Wed, 23 Mar 2011 13:29:52 +0000 (15:29 +0200)]
cfg80211: Add nl80211 event for deletion of a station entry

Indicate an NL80211_CMD_DEL_STATION event when a station entry in
mac80211 is deleted to match with the NL80211_CMD_NEW_STATION event
that is used when the entry was added. This is needed, e.g., to allow
user space to remove a peer from RSN IBSS Authenticator state machine
to avoid re-authentication and re-keying delays when the peer is not
reachable anymore.

Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Support RSN IBSS
Jouni Malinen [Wed, 23 Mar 2011 12:52:19 +0000 (14:52 +0200)]
ath9k: Support RSN IBSS

Add support for using RSN IBSS with ath9k. For now, this uses software
crypto for group addressed frames in RSN IBSS, but that may be
optimized in the future by extending the key cache design to support
per-STA RX GTK.

Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: fix stuck beacon detection
Felix Fietkau [Tue, 22 Mar 2011 20:54:19 +0000 (21:54 +0100)]
ath9k: fix stuck beacon detection

Stuck beacon detection is supposed to trigger when 9 consecutive beacons
could not be sent by the hardware. When the driver runs only one active
AP mode interface, it still configures the hardware beacon timer for
4 (ATH_BCBUF) beacon slots slots, which causes stuck beacon detection
to be reset if ath9k_hw_stoptxdma clears the stuck frames between
SWBA intervals.
Fix this by not resetting the missed beacon count for empty slots and
multiplying the threshold not by the maximum number of beacon slots
but by the configured number of beacon interfaces.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: trigger nfcal only after multiple missed beacons in AP mode
Felix Fietkau [Tue, 22 Mar 2011 20:54:18 +0000 (21:54 +0100)]
ath9k: trigger nfcal only after multiple missed beacons in AP mode

Single missed (i.e. not transmitted) beacons in AP mode are not very rare
and not necessarily an indicator of strong interference, so only trigger
noise floor recalibration when multiple consecutive beacons could not
be transmitted.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: fix beacon timer handling issues
Felix Fietkau [Tue, 22 Mar 2011 20:54:17 +0000 (21:54 +0100)]
ath9k: fix beacon timer handling issues

AP mode beacon timers in ath9k are configured in milliseconds, which breaks
when increasing ATH_BCBUF to 8 instead of 4 (due to rounding errors).
Since the hardware timers are actually configured in microseconds, it's
better to let the driver use that unit directly.

To be able to do that, the beacon interval parameter abuse for passing
certain flags needs to be removed. This is easy to do, because those flags
are completely unnecessary anyway. ATH9K_BEACON_ENA is ignored,
ATH9K_BEACON_RESET_TSF can be replaced with calling ath9k_hw_reset_tsf
from the driver directly.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agolibertas_spi: Add support for suspend/resume
Vasily Khoruzhick [Wed, 16 Mar 2011 14:41:46 +0000 (16:41 +0200)]
libertas_spi: Add support for suspend/resume

Add support for suspend/resume in if_spi.

Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agowireless: mwifiex: initial commit for Marvell mwifiex driver
Bing Zhao [Tue, 22 Mar 2011 01:00:50 +0000 (18:00 -0700)]
wireless: mwifiex: initial commit for Marvell mwifiex driver

This driver adds WiFi support for Marvell 802.11n based chipsets
with SDIO interface. Currently only SD8787 is supported. More
chipsets will be supported later.

drivers/net/wireless/mwifiex/

Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Kiran Divekar <dkiran@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Marc Yang <yangyang@marvell.com>
Signed-off-by: Ramesh Radhakrishnan <rramesh@marvell.com>
Signed-off-by: Frank Huang <frankh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove AR9485 1.0 support
Luis R. Rodriguez [Tue, 22 Mar 2011 00:27:35 +0000 (17:27 -0700)]
ath9k_hw: remove AR9485 1.0 support

Only AR9485 1.1 was sold. This debloats the driver by ~14 KiB.

   text    data     bss     dec     hex filename
 300413     624    1056  302093   49c0d drivers/net/wireless/ath/ath9k/ath9k_hw.ko

   text    data     bss     dec     hex filename
 310285     624    1056  311965   4c29d drivers/net/wireless/ath/ath9k/ath9k_hw-old.ko

$ du -b ath9k_hw*
6210541 ath9k_hw.ko
6225089 ath9k_hw-old.ko

Cc: Bill Wu <bill.wu@atheros.com>
Cc: Paul Shaw <paul.shaw@atheros.com>
Cc: Forbes Tsai <Forbes.Tsai@Atheros.com>
Cc: Jesmine Chen <jesmine.chen@atheros.com>
Cc: Marvian Chen <Hou-hua.Chen@Atheros.com>
Cc: Vivek Natarajan <vivek.natarajan@atheros.com>
Cc: Bernadette Yetso <bernadette.yetso@atheros.com>
Cc: Sarvesh Shrivastava <sarvesh.shrivastava@atheros.com>
Acked-by: Yi-Chen Su <yi-chen.su@atheros.com>
Acked-by: Jeffrey Chung <jeffrey.chung@atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: Minor optimization in tx status handling
Helmut Schaa [Mon, 21 Mar 2011 14:07:55 +0000 (15:07 +0100)]
mac80211: Minor optimization in tx status handling

ieee80211_tx_status iterates over all tx rates the driver reports back
in order to
1) mark tx rates as invalid if the driver cannot have tried that rate
2) find the actually used tx rate for the final retransmission

By leaving the for loop when the first invalid rate index is found we
can move the rates_idx assignment after the loop and therefore save
a few assignments and conditionals.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: remove the pending frames ath_txq_schedule workaround
Felix Fietkau [Sat, 19 Mar 2011 12:55:43 +0000 (13:55 +0100)]
ath9k: remove the pending frames ath_txq_schedule workaround

This workaround called ath_txq_schedule whenever there were still pending
frames for a queue, but the queue depth was zero. Because of its its high
false positive probability (e.g. with paused TIDs) and because it is in
the way of other pending work (AP powersave fixes), it is better to remove
this code entirely.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: force rx chainmask to 7 on AR9100
Felix Fietkau [Sat, 19 Mar 2011 12:55:42 +0000 (13:55 +0100)]
ath9k_hw: force rx chainmask to 7 on AR9100

Most AR9100 devices already have a chainmask of 7 (three antennas), however
on the ones that don't (rx and tx chainmask set to 5), problems with IQ
mismatch calibration have been observed.
This shows up as tx queue hangs (and subsequent hardware resets) if traffic
is sent during this type of calibration.
Forcing the rx chainmask to 7 fixes the calibration issues with no apparent
negative side effects on throughput and stability.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: enable a BlockAck related fixup specific to AR9100
Felix Fietkau [Sat, 19 Mar 2011 12:55:41 +0000 (13:55 +0100)]
ath9k_hw: enable a BlockAck related fixup specific to AR9100

Fixes interop issues with aggregation in combination with multi-BSSID

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: add support for overriding LED pin and GPIO settings from platform data
Felix Fietkau [Sat, 19 Mar 2011 12:55:40 +0000 (13:55 +0100)]
ath9k: add support for overriding LED pin and GPIO settings from platform data

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: add support for overriding the MAC address through platform data
Felix Fietkau [Sat, 19 Mar 2011 12:55:39 +0000 (13:55 +0100)]
ath9k: add support for overriding the MAC address through platform data

On some devices the correct MAC address is not in the EEPROM data, but
stored somewhere else.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: add an interface for overriding the value of specific GPIO pins
Felix Fietkau [Sat, 19 Mar 2011 12:55:38 +0000 (13:55 +0100)]
ath9k: add an interface for overriding the value of specific GPIO pins

Some devices control antenna settings or other things through GPIO pins
of the wireless interface. Add a debugfs interface for changing those
and keeping them set across card resets.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: remove unnecessary debugfs return code checks
Felix Fietkau [Sat, 19 Mar 2011 12:55:37 +0000 (13:55 +0100)]
ath9k: remove unnecessary debugfs return code checks

Since the ath9k debugfs directory is cleaned up by debugfs_remove_recursive,
there's no point in checking the return code of every single debugfs create
line.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agortlwifi: rtl8192c{e,u}: Remove some extraneous casts on memcpy commands
Larry Finger [Fri, 18 Mar 2011 02:53:01 +0000 (21:53 -0500)]
rtlwifi: rtl8192c{e,u}: Remove some extraneous casts on memcpy commands

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: Queue ADDBA requests in respective data queues
Nishant Sarmukadam [Thu, 17 Mar 2011 18:58:49 +0000 (11:58 -0700)]
mwl8k: Queue ADDBA requests in respective data queues

Queue ADDBA requests in respective data queues to avoid ADDBA
requests and the the related data packets (to the same ra/tid)
queued in the hardware to be sent out asynchronously.

Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: Check outgoing rate for a station to decide if ampdu can be created
Nishant Sarmukadam [Thu, 17 Mar 2011 18:58:48 +0000 (11:58 -0700)]
mwl8k: Check outgoing rate for a station to decide if ampdu can be created

If the outgoing packet rate to a particular HT station is <=6.5
Mbps, do not attempt to create an ampdu. Also, if the outgoing
rate is legacy rate, do not create an ampdu.

Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: Handle the watchdog event from the firmware
Nishant Sarmukadam [Thu, 17 Mar 2011 18:58:47 +0000 (11:58 -0700)]
mwl8k: Handle the watchdog event from the firmware

When an ampdu stream is on, if the firmware rate adaptation
logic decides that the outgoing packet rate to the station needs
to go below 6.5Mbps (non HT rate), it sends an event indicating that
the ampdu stream needs to be destroyed. Handle this event in the driver
and destroy the ampdu stream so that the rate can go below 6.5Mbps

Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: Initiate BA sessions
Nishant Sarmukadam [Thu, 17 Mar 2011 18:58:46 +0000 (11:58 -0700)]
mwl8k: Initiate BA sessions

Specifically, handle ampdu_action and attempt to start a BA
session on receiving the first qos packet from mac80211 for
transmission to a HT sta.  While the BA session is being created,
all the packets belonging to that stream will be dropped to
prevent sequence number mismatch at the recipient.

Contains contributions from:
Yogesh Powar <yogeshp@marvell.com>
Pradeep Nemavat <pnemavat@marvell.com>
Brian Cavagnolo <brian@cozybit.com>

Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: add internal API for managing AMPDU streams
Brian Cavagnolo [Thu, 17 Mar 2011 18:58:45 +0000 (11:58 -0700)]
mwl8k: add internal API for managing AMPDU streams

In particular, we can now add, start, lookup, and remove streams.

Based on work by Nishant Sarmukadam <nishants@marvell.com> and
Pradeep Nemavat <pnemavat@marvell.com>.

Signed-off-by: Pradeep Nemavat <pnemavat@marvell.com>
Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: differentiate between WMM queues and AMPDU queues
Brian Cavagnolo [Thu, 17 Mar 2011 18:58:44 +0000 (11:58 -0700)]
mwl8k: differentiate between WMM queues and AMPDU queues

We now have two different kinds of queues.  And the number of
AMPDU queues may vary.  So we must be clear about which queues we
are dealing with.  Note that when we report the number of queues
to mac80211, we only report the WMM queues.

Based on work by Yogesh Powar <yogeshp@marvell.com>.

Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: add support for block ack commands
Nishant Sarmukadam [Thu, 17 Mar 2011 18:58:43 +0000 (11:58 -0700)]
mwl8k: add support for block ack commands

Signed-off-by: Pradeep Nemavat <pnemavat@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: update to ap firmware API version 2
Brian Cavagnolo [Thu, 17 Mar 2011 18:58:42 +0000 (11:58 -0700)]
mwl8k: update to ap firmware API version 2

Firmware APIv2 adds the following enhancements:

-- capabilities are reported by the firmware
-- API supports up to 8 dedicated AMPDU streams
-- optional packet timestamping and expiration can be enabled.
   Specifically, packets that are queued in firmware for longer
   than 500ms will be dropped if this option is used.

Based on work by "Nishant Sarmukadam" <nishants@marvell.com>

Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: refactor in preparation for APIv2 update
Brian Cavagnolo [Thu, 17 Mar 2011 18:58:41 +0000 (11:58 -0700)]
mwl8k: refactor in preparation for APIv2 update

Specifically, APIv2 will specify a variable number of AMPDU
queues in the MWL8K_CMD_GET_HW_SPEC.  So init the tx queues after
MWL8K_CMD_GET_HW_SPEC for ap fw.

Also, we make it safe to deinit queues that have not been init'd.
This happens if the mwl8k_get_hw_spec_ap routine fails, for
example.

Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Brian Cavagnolo <brian@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: add comment about reordering
Johannes Berg [Tue, 15 Mar 2011 22:17:01 +0000 (23:17 +0100)]
mac80211: add comment about reordering

Took me a minute to figure this out, maybe
it's better documented...

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Speedup register ops for HTC driver
Rajkumar Manoharan [Tue, 15 Mar 2011 17:41:35 +0000 (23:11 +0530)]
ath9k_hw: Speedup register ops for HTC driver

Fine-tuning register write operation and avoid unnecessay
delays for ath9k_htc driver, saves hw reset time which
improves scanning time and also solves one of the following
scenario.

Sometimes the ACK is sent by STA for assoc response is not
seen at AP side. So the AP continues to send retry assoc
responses. At the STA side, since the assoc response was
already forwarded to mac80211, it proceeded to channel change
which in turns does chip reset.

In most of the cases the chip reset was completed before
max retries are reached at AP side. Hence STA can able to ACK
the retried frames again. But in clear environment these retries
are completed within shortspan of time.

Since ath9k_htc consumes more time for hw reset, this latency
is causing dissociation by AP due to max reties are reached.
This issue was originally reported with Cisco Aironet 1250 AP
in HT40 mode in noise free environment.

Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath: Speedup key set/reset ops for HTC driver
Rajkumar Manoharan [Tue, 15 Mar 2011 14:25:38 +0000 (19:55 +0530)]
ath: Speedup key set/reset ops for HTC driver

By enabling buffered register write for ath9k_htc driver
avoids unnecessary dissociation while rekeying phase under
heavy traffic exchange.

Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Fix throughput drops in HT40 mode for AR9287 chips
Rajkumar Manoharan [Tue, 15 Mar 2011 14:25:36 +0000 (19:55 +0530)]
ath9k_hw: Fix throughput drops in HT40 mode for AR9287 chips

Doing adc gain calibration for AR9287 chips is causing
throughput drops in HT40 mode. Remove ADC Gain from supported
calibration list.

Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_htc: Fix LED pin for AR9287 HTC device
Rajkumar Manoharan [Tue, 15 Mar 2011 14:25:35 +0000 (19:55 +0530)]
ath9k_htc: Fix LED pin for AR9287 HTC device

Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlwifi: remove duplicate initialization in __iwl_down()
Dan Carpenter [Tue, 15 Mar 2011 07:03:24 +0000 (10:03 +0300)]
iwlwifi: remove duplicate initialization in __iwl_down()

We initialize exit_pending twice.  It's the second initialization which
is correct.  That was added in d745d472af "iwlwifi: cancel scan when
down the device".

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoLinux 2.6.39-rc1
Linus Torvalds [Tue, 29 Mar 2011 19:09:47 +0000 (12:09 -0700)]
Linux 2.6.39-rc1

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc
Linus Torvalds [Tue, 29 Mar 2011 19:09:30 +0000 (12:09 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/cjb/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cjb/mmc: (26 commits)
  mmc: SDHI should depend on SUPERH || ARCH_SHMOBILE
  mmc: tmio_mmc: Move some defines into a shared header
  mmc: tmio: support aggressive clock gating
  mmc: tmio: fix power-mode interpretation
  mmc: tmio: remove work-around for unmasked SDIO interrupts
  sh: fix SDHI IO address-range
  ARM: mach-shmobile: fix SDHI IO address-range
  mmc: tmio: only access registers above 0xff, if available
  mfd: remove now redundant sh_mobile_sdhi.h header
  sh: convert boards to use linux/mmc/sh_mobile_sdhi.h
  ARM: mach-shmobile: convert boards to use linux/mmc/sh_mobile_sdhi.h
  mmc: tmio: convert the SDHI MMC driver from MFD to a platform driver
  sh: ecovec: use the CONFIG_MMC_TMIO symbols instead of MFD
  mmc: tmio: split core functionality, DMA and MFD glue
  mmc: tmio: use PIO for short transfers
  mmc: tmio-mmc: Improve DMA stability on sh-mobile
  mmc: fix mmc_app_send_scr() for dma transfer
  mmc: sdhci-esdhc: enable esdhc on imx53
  mmc: sdhci-esdhc: use writel/readl as general APIs
  mmc: sdhci: add the abort CMDTYPE bits definition
  ...

13 years agoMerge branch 'frv' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux...
Linus Torvalds [Tue, 29 Mar 2011 18:43:30 +0000 (11:43 -0700)]
Merge branch 'frv' of git://git./linux/kernel/git/dhowells/linux-2.6-frv

* 'frv' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-2.6-frv:
  FRV: Use generic show_interrupts()
  FRV: Convert genirq namespace
  frv: Select GENERIC_HARDIRQS_NO_DEPRECATED
  frv: Convert cpu irq_chip to new functions
  frv: Convert mb93493 irq_chip to new functions
  frv: Convert mb93093 irq_chip to new function
  frv: Convert mb93091 irq_chip to new functions
  frv: Fix typo from __do_IRQ overhaul
  frv: Remove stale irq_chip.end
  FRV: Do some cleanups
  FRV: Missing node arg in alloc_thread_info_node() macro
  NOMMU: implement access_remote_vm
  NOMMU: support SMP dynamic percpu_alloc
  NOMMU: percpu should use is_vmalloc_addr().

13 years agoMerge branch 'stable/bug-fixes-rc1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 29 Mar 2011 18:36:52 +0000 (11:36 -0700)]
Merge branch 'stable/bug-fixes-rc1' of git://git./linux/kernel/git/konrad/xen

* 'stable/bug-fixes-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen: Use new irq_move functions
  xen: Convert genirq namespace
  xen: fix p2m section mismatches
  xen/p2m: Allocate p2m tracking pages on override
  xen-gntdev: unlock on error path in gntdev_mmap()
  xen-gntdev: return -EFAULT on copy_to_user failure

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Tue, 29 Mar 2011 18:20:09 +0000 (11:20 -0700)]
Merge git://git./linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  watchdog: softdog.c: enhancement to optionally invoke panic instead of reboot on timer expiry
  watchdog: fix nv_tco section mismatch
  watchdog: sp5100_tco.c: Check if firmware has set correct value in tcobase.
  watchdog: Convert release_resource to release_region/release_mem_region
  watchdog: s3c2410_wdt.c: Convert release_resource to release_region/release_mem_region

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Linus Torvalds [Tue, 29 Mar 2011 18:11:18 +0000 (11:11 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/bp/bp

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
  amd64_edac: Fix potential memleak

13 years agoMerge branch 'irq-final-for-linus-v2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 29 Mar 2011 17:46:15 +0000 (10:46 -0700)]
Merge branch 'irq-final-for-linus-v2' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'irq-final-for-linus-v2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (111 commits)
  gpio: ab8500: Mark broken
  genirq: Remove move_*irq leftovers
  genirq: Remove compat code
  drivers: Final irq namespace conversion
  mn10300: Use generic show_interrupts()
  mn10300: Cleanup irq_desc access
  mn10300: Convert genirq namespace
  frv: Use generic show_interrupts()
  frv: Convert genirq namespace
  frv: Select GENERIC_HARDIRQS_NO_DEPRECATED
  frv: Convert cpu irq_chip to new functions
  frv: Convert mb93493 irq_chip to new functions
  frv: Convert mb93093 irq_chip to new function
  frv: Convert mb93091 irq_chip to new functions
  frv: Fix typo from __do_IRQ overhaul
  frv: Remove stale irq_chip.end
  m68k: Convert irq function namespace
  xen: Use new irq_move functions
  xen: Cleanup genirq namespace
  unicore32: Use generic show_interrupts()
  ...

13 years agochar/tpm: Fix unitialized usage of data buffer
Peter Huewe [Tue, 29 Mar 2011 11:31:25 +0000 (13:31 +0200)]
char/tpm: Fix unitialized usage of data buffer

This patch fixes information leakage to the userspace by initializing
the data buffer to zero.

Reported-by: Peter Huewe <huewe.external@infineon.com>
Signed-off-by: Peter Huewe <huewe.external@infineon.com>
Signed-off-by: Marcel Selhorst <m.selhorst@sirrix.com>
[ Also removed the silly "* sizeof(u8)".  If that isn't 1, we have way
  deeper problems than a simple multiplication can fix.   - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoamd64_edac: Fix potential memleak
Borislav Petkov [Tue, 29 Mar 2011 16:10:53 +0000 (18:10 +0200)]
amd64_edac: Fix potential memleak

We check the pointers together but at least one of them could be invalid
due to failed allocation. Since we cannot continue if either of the two
allocations has failed, exit early by freeing them both.

Cc: <stable@kernel.org> # 38.x
Reported-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agofs: don't use igrab() while holding i_lock
Dave Chinner [Tue, 29 Mar 2011 07:08:50 +0000 (18:08 +1100)]
fs: don't use igrab() while holding i_lock

Fix the incorrect use of igrab() inside the i_lock in NFS and Ceph‥

If we are already holding the i_lock, we have a reference to the
inode so we can safely use ihold() to gain an extra reference. This
avoids hangs due to lock recursion on the i_lock now that the
inode_lock is gone and igrab() uses the i_lock itself.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-fsdevel@vger.kernel.org
Cc: Ryan Mallon <ryan@bluewatersys.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 29 Mar 2011 14:41:33 +0000 (07:41 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (30 commits)
  xfrm: Restrict extended sequence numbers to esp
  xfrm: Check for esn buffer len in xfrm_new_ae
  xfrm: Assign esn pointers when cloning a state
  xfrm: Move the test on replay window size into the replay check functions
  netdev: bfin_mac: document TE setting in RMII modes
  drivers net: Fix declaration ordering in inline functions.
  cxgb3: Apply interrupt coalescing settings to all queues
  net: Always allocate at least 16 skb frags regardless of page size
  ipv4: Don't ip_rt_put() an error pointer in RAW sockets.
  net: fix ethtool->set_flags not intended -EINVAL return value
  mlx4_en: Fix loss of promiscuity
  tg3: Fix inline keyword usage
  tg3: use <linux/io.h> and <linux/uaccess.h> instead <asm/io.h> and <asm/uaccess.h>
  net: use CHECKSUM_NONE instead of magic number
  Net / jme: Do not use legacy PCI power management
  myri10ge: small rx_done refactoring
  bridge: notify applications if address of bridge device changes
  ipv4: Fix IP timestamp option (IPOPT_TS_PRESPEC) handling in ip_options_echo()
  can: c_can: Fix tx_bytes accounting
  can: c_can_platform: fix irq check in probe
  ...

13 years agoxen: Use new irq_move functions
Thomas Gleixner [Thu, 24 Mar 2011 20:37:31 +0000 (21:37 +0100)]
xen: Use new irq_move functions

These functions take irq_data as an argument and avoid a redundant
lookup in the sparse irq case.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
13 years agoxen: Convert genirq namespace
Thomas Gleixner [Fri, 25 Mar 2011 10:01:39 +0000 (11:01 +0100)]
xen: Convert genirq namespace

Converted with coccinelle.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
13 years agoxen: fix p2m section mismatches
Randy Dunlap [Thu, 24 Mar 2011 20:34:32 +0000 (13:34 -0700)]
xen: fix p2m section mismatches

Fix section mismatch warnings:
set_phys_range_identity() is called by __init xen_set_identity(),
so also mark set_phys_range_identity() as __init.
then:
__early_alloc_p2m() is called set_phys_range_identity(), so also mark
__early_alloc_p2m() as __init.

WARNING: arch/x86/built-in.o(.text+0x7856): Section mismatch in reference from the function __early_alloc_p2m() to the function .init.text:extend_brk()
The function __early_alloc_p2m() references
the function __init extend_brk().
This is often because __early_alloc_p2m lacks a __init
annotation or the annotation of extend_brk is wrong.

WARNING: arch/x86/built-in.o(.text+0x7967): Section mismatch in reference from the function set_phys_range_identity() to the function .init.text:extend_brk()
The function set_phys_range_identity() references
the function __init extend_brk().
This is often because set_phys_range_identity lacks a __init
annotation or the annotation of extend_brk is wrong.

[v2: Per Stephen Hemming recommonedation made __early_alloc_p2m static]
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
13 years agoFRV: Use generic show_interrupts()
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
FRV: Use generic show_interrupts()

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agoFRV: Convert genirq namespace
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
FRV: Convert genirq namespace

Convert to new function names.  Converted with coccinelle.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Select GENERIC_HARDIRQS_NO_DEPRECATED
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
frv: Select GENERIC_HARDIRQS_NO_DEPRECATED

All chips converted

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Convert cpu irq_chip to new functions
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
frv: Convert cpu irq_chip to new functions

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Convert mb93493 irq_chip to new functions
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
frv: Convert mb93493 irq_chip to new functions

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Convert mb93093 irq_chip to new function
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
frv: Convert mb93093 irq_chip to new function

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Convert mb93091 irq_chip to new functions
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
frv: Convert mb93091 irq_chip to new functions

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Fix typo from __do_IRQ overhaul
Thomas Gleixner [Tue, 29 Mar 2011 13:05:13 +0000 (14:05 +0100)]
frv: Fix typo from __do_IRQ overhaul

Compiles way better.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agofrv: Remove stale irq_chip.end
Thomas Gleixner [Tue, 29 Mar 2011 13:05:12 +0000 (14:05 +0100)]
frv: Remove stale irq_chip.end

irq_chip.end got obsolete with the removal of __do_IRQ().

irq-mb93093.c even lacks an implementation, but nobody noticed that
it's broken since commit 88d6e1 in 2006.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agoFRV: Do some cleanups
Amerigo Wang [Tue, 29 Mar 2011 13:05:12 +0000 (14:05 +0100)]
FRV: Do some cleanups

1. frv doesn't support SMP, remove the useless SMP bits.

2. frv has its own alloc_task_struct, so define __HAVE_ARCH_TASK_STRUCT_ALLOCATOR
   (I am not sure if frv should use generic alloc_task_struct().)

Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>
13 years agoFRV: Missing node arg in alloc_thread_info_node() macro
David Howells [Tue, 29 Mar 2011 13:05:12 +0000 (14:05 +0100)]
FRV: Missing node arg in alloc_thread_info_node() macro

There are two alloc_thread_info_node() macros defined (one for debugging and
one for normal).  The commit that changed them most recently:

commit b6a84016bd2598e35ead635147fa53619982648d
Author: Eric Dumazet <eric.dumazet@gmail.com>
Date:   Tue Mar 22 16:30:42 2011 -0700
Subject: mm: NUMA aware alloc_thread_info_node()

didn't add the node argument into the macro argument list for the normal macro.
This results in the following error:

kernel/fork.c:267:39: error: macro "alloc_thread_info_node" passed 2 arguments, but takes just 1
kernel/fork.c: In function 'dup_task_struct':
kernel/fork.c:267: error: 'alloc_thread_info_node' undeclared (first use in this function)
kernel/fork.c:267: error: (Each undeclared identifier is reported only once
kernel/fork.c:267: error: for each function it appears in.)

Signed-off-by: David Howells <dhowells@redhat.com>
13 years agoNOMMU: implement access_remote_vm
Mike Frysinger [Tue, 29 Mar 2011 13:05:12 +0000 (14:05 +0100)]
NOMMU: implement access_remote_vm

Recent vm changes brought in a new function which the core procfs code
utilizes.  So implement it for nommu systems too to avoid link failures.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Simon Horman <horms@verge.net.au>
Tested-by: Ithamar Adema <ithamar.adema@team-embedded.nl>
Acked-by: Greg Ungerer <gerg@uclinux.org>
13 years agogpio: ab8500: Mark broken
Thomas Gleixner [Tue, 29 Mar 2011 02:02:50 +0000 (04:02 +0200)]
gpio: ab8500: Mark broken

This driver is broken in several aspects.

 1) old style irq_chip functions. Sigh

 2) Abuse of the unlock callback. That's not supposed to be a state
    machine for evrything and some more.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
13 years agogenirq: Remove move_*irq leftovers
Thomas Gleixner [Tue, 29 Mar 2011 00:51:13 +0000 (02:51 +0200)]
genirq: Remove move_*irq leftovers

All users converted to new interface.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
13 years agogenirq: Remove compat code
Thomas Gleixner [Mon, 28 Mar 2011 11:32:20 +0000 (13:32 +0200)]
genirq: Remove compat code

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
13 years agodrivers: Final irq namespace conversion
Thomas Gleixner [Mon, 28 Mar 2011 15:49:12 +0000 (17:49 +0200)]
drivers: Final irq namespace conversion

Scripted with coccinelle.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
13 years agomn10300: Use generic show_interrupts()
Thomas Gleixner [Thu, 24 Mar 2011 17:54:24 +0000 (18:54 +0100)]
mn10300: Use generic show_interrupts()

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
13 years agomn10300: Cleanup irq_desc access
Thomas Gleixner [Thu, 24 Mar 2011 16:36:37 +0000 (17:36 +0100)]
mn10300: Cleanup irq_desc access

The migration needs only access to irq_data.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
13 years agomn10300: Convert genirq namespace
Thomas Gleixner [Thu, 24 Mar 2011 16:35:56 +0000 (17:35 +0100)]
mn10300: Convert genirq namespace

Convert to new function names. Converted with coccinelle.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>