GitHub/mt8127/android_kernel_alcatel_ttab.git
16 years agomac80211: fix deadlock in sta->lock
Tomas Winkler [Tue, 27 May 2008 14:50:52 +0000 (17:50 +0300)]
mac80211: fix deadlock in sta->lock

This patch fixes a deadlock of sta->lock use, occurring while changing
tx aggregation states, as dev_queue_xmit end up in new function
test_and_clear_sta_flags that uses that lock thus leading to deadlock

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: fix ieee80211_get_buffered_bc
Tomas Winkler [Tue, 27 May 2008 14:50:51 +0000 (17:50 +0300)]
mac80211: fix ieee80211_get_buffered_bc

fix bss not initialized in ieee80211_get_buffered_bc
and unbalanced locking

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agorndis_wlan: use ARRAY_SIZE instead of sizeof when adding 11g rates
Scott Ashcroft [Tue, 27 May 2008 08:15:02 +0000 (11:15 +0300)]
rndis_wlan: use ARRAY_SIZE instead of sizeof when adding 11g rates

While figuring out the TKIP problem I found a bug in the code which adds the 11g
rates. It's using sizeof instead of ARRAY_SIZE to terminate the for loop. This makes
it fall off the end of the rates array start into the frequency array instead. Running
"iwlist rate" should show the problem as there will always be 32 rates with the last
few being bogus.

The following patch will fix it.

Signed-off-by: Scott Ashcroft <scott.ashcroft@talk21.com>
Signed-off-by: Jussi Kivilinna <jussi.kivilinna@mbnet.fi>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoath5k: use IEEE80211_SKB_CB
Johannes Berg [Mon, 26 May 2008 14:43:39 +0000 (16:43 +0200)]
ath5k: use IEEE80211_SKB_CB

ath5k still uses the "(void*) skb->cb" direct cast, use IEEE80211_SKB_CB
instead.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: before sleeping, check for a command result
Holger Schurig [Fri, 23 May 2008 14:04:13 +0000 (16:04 +0200)]
libertas: before sleeping, check for a command result

If we don't check for a command response early, but rather sleep,
then we might sleep despite an already-received command response.
This will lead to a command-timeout.

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: use lbs_pr_XX instead of printk
Holger Schurig [Fri, 23 May 2008 10:16:51 +0000 (12:16 +0200)]
libertas: use lbs_pr_XX instead of printk

... because lbs_pr_XXX prefixes all messages with "libertas: "

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: speeds up downloading of CF firmware
Holger Schurig [Fri, 23 May 2008 08:18:26 +0000 (10:18 +0200)]
libertas: speeds up downloading of CF firmware

Keep the timeout the same (1000*500 == 100000 * 5), but take shorter
naps. Makes downloading the firmware slightly faster.

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: don't spin_unlock_irq() twice
Holger Schurig [Fri, 23 May 2008 08:07:56 +0000 (10:07 +0200)]
libertas: don't spin_unlock_irq() twice

priv->driver_lock has already been unlocked some lines above. This patch
fixes the sparse warning:

drivers/net/wireless/libertas/main.c:792:6: warning: context problem in 'lbs_thread': '_spin_unlock_irq' expected different context
drivers/net/wireless/libertas/main.c:792:6:    context 'lock': wanted >= 1, got 0

Signed-of-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: clean up skb reallocation code
Johannes Berg [Thu, 29 May 2008 08:38:53 +0000 (10:38 +0200)]
mac80211: clean up skb reallocation code

This cleans up the skb reallocation code to avoid problems with
skb->truesize, not resize an skb twice for a single output path
because we didn't expand it enough during the first copy and also
removes the code to further expand it during crypto operations
which will no longer be necessary.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agops3: gelic: updown_lock semaphore to mutex
Daniel Walker [Thu, 22 May 2008 07:00:03 +0000 (00:00 -0700)]
ps3: gelic: updown_lock semaphore to mutex

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Acked-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agops3: gelic: assoc_stat_lock semaphore to mutex
Daniel Walker [Thu, 22 May 2008 07:00:02 +0000 (00:00 -0700)]
ps3: gelic: assoc_stat_lock semaphore to mutex

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Acked-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agops3: gelic: scan_lock semaphore to mutex
Daniel Walker [Thu, 22 May 2008 07:00:01 +0000 (00:00 -0700)]
ps3: gelic: scan_lock semaphore to mutex

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Acked-by: Masakazu Mokuno <mokuno@sm.sony.co.jp>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: provide reset_card() callback on OLPC
David Woodhouse [Tue, 20 May 2008 15:48:52 +0000 (16:48 +0100)]
libertas: provide reset_card() callback on OLPC

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: enable mesh
Johannes Berg [Tue, 20 May 2008 10:10:49 +0000 (12:10 +0200)]
b43: enable mesh

This patch enables b43 to do mesh networking, tested against my zd1211rw
dongle.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Reviewed-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: remove channel use statistics
Johannes Berg [Wed, 21 May 2008 15:33:42 +0000 (17:33 +0200)]
mac80211: remove channel use statistics

The useless channel use statistics are quite a lot of code, currently
use integer divisions in the packet fast path, are rather inaccurate
since they do not account for retries and finally nobody even cares.
Hence, remove them completely.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: Add firmware markers support
Michael Buesch [Mon, 19 May 2008 22:24:36 +0000 (00:24 +0200)]
b43: Add firmware markers support

This adds support for firmware markers.
With firmware markers it's easily possible to check whether the
firmware runs some codepath or not. The driver will throw a message
when the firmware executes a MARKER(x).

Signed-off-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: Add panic reason code that doesn't trigger restart
Michael Buesch [Mon, 19 May 2008 21:51:37 +0000 (23:51 +0200)]
b43: Add panic reason code that doesn't trigger restart

Add a firmware panic reason code that doesn't trigger a restart.
This is useful for firmware debugging and avoiding endless
restart loops. We can use FWPANIC_DIE to halt the firmware at a
well defined point.

Signed-off-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: Allow running without PCM firmware
Michael Buesch [Sat, 17 May 2008 21:43:57 +0000 (23:43 +0200)]
b43: Allow running without PCM firmware

This patch adds code to allow running the device without PCM firmware loaded.
Without PCM firmware we don't have hardware accelerated crypto on
devices with a core rev <= 10.

Signed-off-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: Add hooks for firmware debugging
Michael Buesch [Sat, 17 May 2008 20:44:35 +0000 (22:44 +0200)]
b43: Add hooks for firmware debugging

This patch adds some hooks for firmware debugging.

Signed-off-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: use multi-queue master netdevice
Johannes Berg [Fri, 16 May 2008 22:57:14 +0000 (00:57 +0200)]
mac80211: use multi-queue master netdevice

This patch updates mac80211 and drivers to be multi-queue aware and
use that instead of the internal queue mapping. Also does a number
of cleanups in various pieces of the code that fall out and reduces
internal mac80211 state size.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: dont allow fragmentation and requeuing on A-MPDU queues
Johannes Berg [Fri, 16 May 2008 22:57:13 +0000 (00:57 +0200)]
mac80211: dont allow fragmentation and requeuing on A-MPDU queues

There really is no reason for a driver to reject a frame on
an A-MPDU queue when it can stop that queue for any period
of time and is given frames one by one. Hence, disallow it
with a big warning and reduce mac80211-internal state.

Also add a warning when we try to fragment a frame destined
for an A-MPDU queue and drop it, the actual bug needs to be
fixed elsewhere but I'm not exactly sure how to yet.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agob43: Fix typo in firmware file name for 802.11 cores with rev 13
Larry.Finger@lwfinger.net [Thu, 15 May 2008 19:07:36 +0000 (14:07 -0500)]
b43: Fix typo in firmware file name for 802.11 cores with rev 13

When the patch for the BCM4311 rev 2 was prepared, I misread the specs
and coded the wrong file name for the initvals firmware.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agozd1211rw: Use DMA-aware buffer for usb transfer
Atsushi Nemoto [Fri, 16 May 2008 08:27:05 +0000 (17:27 +0900)]
zd1211rw: Use DMA-aware buffer for usb transfer

The zd1211rw driver uses unaligned stack buffer for USB control
message.  But it might cause stack corruption on non-coherent
platform, such as MIPS.  Use DMA-aware buffers for USB transfer.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: move TX info into skb->cb
Johannes Berg [Thu, 15 May 2008 10:55:29 +0000 (12:55 +0200)]
mac80211: move TX info into skb->cb

This patch converts mac80211 and all drivers to have transmit
information and status in skb->cb rather than allocating extra
memory for it and copying all the data around. To make it fit,
a union is used where only data that is necessary for all steps
is kept outside of the union.

A number of fixes were done by Ivo, as well as the rt2x00 part
of this patch.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: reorder some transmit handlers
Johannes Berg [Thu, 15 May 2008 10:55:28 +0000 (12:55 +0200)]
mac80211: reorder some transmit handlers

The next patch will require that transmit handlers that are after
fragmentation are aware of the fact that the control info is also
fragmented. To make that easier, this patch moves a number of
transmit handlers before fragmentation.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: use rate index in TX control
Johannes Berg [Thu, 15 May 2008 10:55:27 +0000 (12:55 +0200)]
mac80211: use rate index in TX control

This patch modifies struct ieee80211_tx_control to give band
info and the rate index (instead of rate pointers) to drivers.
This mostly serves to reduce the TX control structure size to
make it fit into skb->cb so that the fragmentation code can
put it there and we can think about passing it to drivers that
way in the future.

The rt2x00 driver update was done by Ivo, thanks.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: let drivers wake but not start queues
Johannes Berg [Thu, 15 May 2008 10:55:26 +0000 (12:55 +0200)]
mac80211: let drivers wake but not start queues

Having drivers start queues is just confusing, their ->start()
callback can block and do whatever is necessary, so let mac80211
start queues and have drivers wake queues when necessary (to get
packets flowing again right away.)

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: fix bugs in queue handling functions
Johannes Berg [Thu, 15 May 2008 10:55:25 +0000 (12:55 +0200)]
mac80211: fix bugs in queue handling functions

Commit 55c308c1315bc7267dbb88011c208fd743cdce31
("mac80211: QoS related cleanups") introduced another bug,
the queue handling functions that operate on all queues now
only operated on the first queues, not the A-MPDU queues as
expected. This patch fixes this.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: trigger event log from debugfs
Ester Kummer [Thu, 15 May 2008 05:54:18 +0000 (13:54 +0800)]
iwlwifi: trigger event log from debugfs

This patch adds a trigger for event log printing to debugfs.
It removes the triger from sysfs.

Signed-off-by: Ester Kummer <ester.kummer@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: refactor pci prob flow
Tomas Winkler [Thu, 15 May 2008 05:54:17 +0000 (13:54 +0800)]
iwlwifi: refactor pci prob flow

This patch refactores pci prob flow. It moves mac80211 registration
to the end, otherwise there is a race between error path in pci_probe
and mac_start.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: add debugfs to disable/enable run time calibration
Tomas Winkler [Thu, 15 May 2008 05:54:16 +0000 (13:54 +0800)]
iwlwifi: add debugfs to disable/enable run time calibration

This patch adds functionality to debugfs to enable or disable chain
noise or sensitivity calibrations.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: remove 4965 remainings in iwl-eeprom.c file
Tomas Winkler [Thu, 15 May 2008 05:54:15 +0000 (13:54 +0800)]
iwlwifi: remove 4965 remainings in iwl-eeprom.c file

The patch cleans up 4965 remaings from iwl-eeprom.c file.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: iwl5000 WiFi/WiMax coexistence
Ron Rindjunsky [Thu, 15 May 2008 05:54:14 +0000 (13:54 +0800)]
iwlwifi: iwl5000 WiFi/WiMax coexistence

This patch adds WiFi/WiMax coexistence to iwl5000.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: add ucode init flow handling for iwl5000
Ron Rindjunsky [Thu, 15 May 2008 05:54:13 +0000 (13:54 +0800)]
iwlwifi: add ucode init flow handling for iwl5000

This patch adds all the handlers and functions needed for ucode
initialization flow.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: add ucode loaders for iwl5000
Ron Rindjunsky [Thu, 15 May 2008 05:54:12 +0000 (13:54 +0800)]
iwlwifi: add ucode loaders for iwl5000

This patch adds ucode initialization handler and functions to load
ucode for iwl5000 NIC.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: add rx_handlers stub for iwl5000
Ron Rindjunsky [Thu, 15 May 2008 05:54:11 +0000 (13:54 +0800)]
iwlwifi: add rx_handlers stub for iwl5000

This patch adds rx_handler_setup stub to iwl5000.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: rename and move Tx queue activation/deactivation
Ron Rindjunsky [Thu, 15 May 2008 05:54:10 +0000 (13:54 +0800)]
iwlwifi: rename and move Tx queue activation/deactivation

This patch moves iwl4965_txq_ctx_activate and iwl4965_txq_ctx_deactivate
to iwl-dev.h and removes 4965 prefix from their names.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: clean up and bug fix for security
Emmanuel Grumbach [Thu, 15 May 2008 05:54:09 +0000 (13:54 +0800)]
iwlwifi: clean up and bug fix for security

This patch cleans up code in security.

1) uses the new pointer to ieee80211_key_conf passed with the tx_control.
2) resolves bug reported by Mirco Tischler (sends ADD_STA in ASYNC mode)
3) resolves bug reported by Volker Braun regarding dynamic WEP
4) drops a WEP packet which has been garbaged by firmware. This can
happen upon rekeying.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: don't switch to SGI if not supported by AP
Emmanuel Grumbach [Thu, 15 May 2008 05:54:08 +0000 (13:54 +0800)]
iwlwifi: don't switch to SGI if not supported by AP

This patch fixes SGI support. RS didn't look at the capabilities of the AP
before switching to SGI, this should lead to a stall in the traffic with an
AP that doesn't support SGI.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Guy Cohen <guy.cohen@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: move TX code into iwl-tx.c
Tomas Winkler [Thu, 15 May 2008 05:54:07 +0000 (13:54 +0800)]
iwlwifi: move TX code into iwl-tx.c

This patch moves the sending part of the TX code into iwl-tx.c
including sending host commands.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: refactor ieee80211_get_qos_ctrl
Tomas Winkler [Thu, 15 May 2008 05:54:06 +0000 (13:54 +0800)]
iwlwifi: refactor ieee80211_get_qos_ctrl

This patch refactors ieee80211_get_qos_ctrl function and its usage
in iwlwifi drivers. Function is moved as inline into iwl-helpers.h.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: remove 4956 form iwl4965_tx_cmd
Tomas Winkler [Thu, 15 May 2008 05:54:05 +0000 (13:54 +0800)]
iwlwifi: remove 4956 form iwl4965_tx_cmd

This patch renames iwl4965_tx_cmd to iwl_tx_cmd and cleans
collateral code.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: move more station managment into iwl-sta.c
Tomas Winkler [Thu, 15 May 2008 05:54:04 +0000 (13:54 +0800)]
iwlwifi: move more station managment into iwl-sta.c

This patch moves station management functions into iwl-sta.c (iwlcore).

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: move iwl_bcast_addr to iwlcore
Tomas Winkler [Thu, 15 May 2008 05:54:03 +0000 (13:54 +0800)]
iwlwifi: move iwl_bcast_addr to iwlcore

This patch renames iwl4965_broadcast_addr to iwl_bcast_addr
and moves it to iwlcore.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: remove 4965 from iwl4965_rate_info
Tomas Winkler [Thu, 15 May 2008 05:54:02 +0000 (13:54 +0800)]
iwlwifi: remove 4965 from iwl4965_rate_info

This patch removes 4965 from iwl4965_rate_info structure and
associated variables.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: remove 4965 prefix from iwl4965_frame
Tomas Winkler [Thu, 15 May 2008 05:54:01 +0000 (13:54 +0800)]
iwlwifi: remove 4965 prefix from iwl4965_frame

This patch removes 4965 from iwl4965_frame struct and handling functions.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: remove 4965 from iwl4965_tx_queue_update_write_ptr
Tomas Winkler [Thu, 15 May 2008 05:54:00 +0000 (13:54 +0800)]
iwlwifi: remove 4965 from iwl4965_tx_queue_update_write_ptr

This patch renames iwl4965_tx_queue_update_write_ptr to
iwl_txq_update_write_ptr. This is a preparation for moving to iwl-tx.c.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: get_hw_cmd_size
Gregory Greenman [Thu, 15 May 2008 05:53:59 +0000 (13:53 +0800)]
iwlwifi: get_hw_cmd_size

This patch introduces a new handler get_hw_cmd_size in hcmd_utils,
which should adjust the size of the command sent to the microcode
according to the current nic.
It also changes the RXON flow to make usage of this new handler.

The patch also adds iwl_rxon_cmd structure which is supperset for
5000 HW and 4965.

Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: rename iwl4965_tx_info to iwl_tx_info
Tomas Winkler [Thu, 15 May 2008 05:53:58 +0000 (13:53 +0800)]
iwlwifi: rename iwl4965_tx_info to iwl_tx_info

This patch renames iwl4965_tx_info to iwl_tx_info.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: rename iwl4965_queue to iwl_queue
Tomas Winkler [Thu, 15 May 2008 05:53:57 +0000 (13:53 +0800)]
iwlwifi: rename iwl4965_queue to iwl_queue

This patch renames iwl4965_queue to iwl_queue.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: filling Tx MCS set
Ron Rindjunsky [Thu, 15 May 2008 05:53:56 +0000 (13:53 +0800)]
iwlwifi: filling Tx MCS set

This patch fills the needed data about HW capabilities in matters of
possible HT Tx MCS.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: separate Tx and Rx MCS when configuring HT
Ron Rindjunsky [Thu, 15 May 2008 05:53:55 +0000 (13:53 +0800)]
mac80211: separate Tx and Rx MCS when configuring HT

This patch follows the 11n spec in separation between Tx and Rx MCS
capabilities. Up until now, when configuring the HT possible set of Tx
MCS only Rx MCS were considered, assuming they are the same as the Tx MCS.
This patch fixed this by looking at low level driver Tx capabilities.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: remove iwl4965_nic_start function
Ron Rindjunsky [Thu, 15 May 2008 05:53:54 +0000 (13:53 +0800)]
iwlwifi: remove iwl4965_nic_start function

This patch moves the contant of iwl4965_nic_start to the only place it is
being used.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: changing irrelevant comment
Ron Rindjunsky [Thu, 15 May 2008 05:53:53 +0000 (13:53 +0800)]
iwlwifi: changing irrelevant comment

This patch changes an old eeprom comment in probe flow.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: tkip.c use struct tkip_ctx in phase 2 key mixing
Harvey Harrison [Wed, 14 May 2008 23:26:20 +0000 (16:26 -0700)]
mac80211: tkip.c use struct tkip_ctx in phase 2 key mixing

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: tkip.c use struct tkip_ctx in phase 1 key mixing
Harvey Harrison [Wed, 14 May 2008 23:26:19 +0000 (16:26 -0700)]
mac80211: tkip.c use struct tkip_ctx in phase 1 key mixing

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: add a struct to hold tkip context
Harvey Harrison [Wed, 14 May 2008 23:26:19 +0000 (16:26 -0700)]
mac80211: add a struct to hold tkip context

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: add const, remove unused function, make one function static
Harvey Harrison [Wed, 14 May 2008 23:26:18 +0000 (16:26 -0700)]
mac80211: add const, remove unused function, make one function static

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: introduce struct michael_mic_ctx and static helpers
Harvey Harrison [Wed, 14 May 2008 23:26:17 +0000 (16:26 -0700)]
mac80211: introduce struct michael_mic_ctx and static helpers

Replace the existing macro with a static function, significantly shrinks the
size of the produced object file.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: michael.c use kernel-provided infrastructure
Harvey Harrison [Wed, 14 May 2008 23:26:16 +0000 (16:26 -0700)]
mac80211: michael.c use kernel-provided infrastructure

Replace private implementation of bit rotation and unaligned access helpers
with kernel-provided implementation.

Fold xswap helper in its one usage in the michael_block macro.

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Cc: "John W. Linville" <linville@tuxdriver.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: reduce command retry time
Holger Schurig [Wed, 14 May 2008 14:30:28 +0000 (16:30 +0200)]
libertas: reduce command retry time

[PATCH, take 2] libertas: reduce command retry time

In the normal case, an unsuccessful command would be retried for 10*5 seconds,
or 10*10 seconds in the worst case. This patch reduces this to 3*3 seconds,
or 3*10 seconds in the worst case.

I also reduced the time it takes to start a new command downloaded.

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Consolidate hash kfree-ing in mesh.c.
Pavel Emelyanov [Wed, 7 May 2008 15:57:11 +0000 (19:57 +0400)]
mac80211: Consolidate hash kfree-ing in mesh.c.

There are already two places, that kfree the mesh_table and
its buckets.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Merge error paths in mesh_table_grow().
Pavel Emelyanov [Wed, 7 May 2008 15:55:59 +0000 (19:55 +0400)]
mac80211: Merge error paths in mesh_table_grow().

This is the first (of two) clean ups after the fixes above.

The err variable is not even required after this cleaning.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Do not report false error from mesh_path_add.
Pavel Emelyanov [Wed, 7 May 2008 15:54:04 +0000 (19:54 +0400)]
mac80211: Do not report false error from mesh_path_add.

In case the hash grow failed, it is not fair to return error -
the new node _was_ _actually_ added in this case.

Besides, after my previous patch, this grow is more likely
to fail on large hashes.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Fix sleeping allocation under lock in mesh_path_node_copy.
Pavel Emelyanov [Wed, 7 May 2008 15:51:51 +0000 (19:51 +0400)]
mac80211: Fix sleeping allocation under lock in mesh_path_node_copy.

The mesh_path_node_copy() can be called like this:
mesh_path_add
 `- write_lock(&pathtbl_resize_lock); /* ! */
 `- mesh_table_grow
     `- ->copy_node
           `- mesh_path_node_copy

thus, the GFP_KERNEL is not suitable here.

The acceptable fix, I suppose, is make this allocation GPF_ATOMIC -
the mpath_node being allocated is 4 pointers, i.e. this allocation
is small enough to survive even under a moderate memory pressure.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Report allocation failure from mesh_path_node_copy.
Pavel Emelyanov [Wed, 7 May 2008 15:48:14 +0000 (19:48 +0400)]
mac80211: Report allocation failure from mesh_path_node_copy.

Now - return the -ENOMEM in case kmalloc fails.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Prepare mesh_table_grow to failing copy_node callback.
Pavel Emelyanov [Wed, 7 May 2008 15:47:01 +0000 (19:47 +0400)]
mac80211: Prepare mesh_table_grow to failing copy_node callback.

The mesh_path_node_copy() performs kmalloc() and thus - may fail
(well, it does not now, but I'm fixing this right now). Its caller -
the mesh_table_grow() - isn't prepared for such a trick yet.

This preparation is just flush the new hash and make copy_node()
return an int value.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Fix one more call to synchronize_rcu in atomic context.
Pavel Emelyanov [Wed, 7 May 2008 15:44:20 +0000 (19:44 +0400)]
mac80211: Fix one more call to synchronize_rcu in atomic context.

(This set applies OK without the previous one of 4 patches,
 but with some fuzz in the 7th one)

The mesh_path_node_free() does so under hashwlock.

But, this one is called
1. from mesh_path_add() after an old hash is hidden and
   synchronize_rcu() is calld
2. mesh_pathtbl_unregister(), when the module is being
   unloaded and no devices exist to mess with this hash.

So, it seems to me, that simply removing the call is OK.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: Brush up error paths in mesh_path_add.
Pavel Emelyanov [Tue, 6 May 2008 14:53:43 +0000 (18:53 +0400)]
mac80211: Brush up error paths in mesh_path_add.

There are already tree paths, that do incremental rollbacks, so
merge them together, rename labels and format the code to look a
bit nicer.

(I do not mind dropping/delaying this patch however).

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: sysfs interface for accessing default mesh channel
Javier Cardona [Tue, 20 May 2008 22:18:49 +0000 (15:18 -0700)]
libertas: sysfs interface for accessing default mesh channel

This will create the following entry:
/sys/class/net/mshX
            -- boot_options
            |   |-- ...
            |   `-- channel
            ...

... which I overlooked on my previous patch.

Signed-off-by: Javier Cardona <javier@cozybit.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: Add reset_card() callback to hardware driver
David Woodhouse [Tue, 20 May 2008 15:43:31 +0000 (16:43 +0100)]
libertas: Add reset_card() callback to hardware driver

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: sysfs interface for accessing non-volatile configuration
Javier Cardona [Sun, 18 May 2008 04:01:24 +0000 (21:01 -0700)]
libertas: sysfs interface for accessing non-volatile configuration

This will create the following sysfs directories:
/sys/class/net/mshX
...
|-- boot_options
|   |-- bootflag
|   `-- boottime
...
|-- mesh_ie
|   |-- capability
|   |-- mesh_id
|   |-- metric_id
|   `-- protocol_id

Signed-off-by: Javier Cardona <javier@cozybit.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: Extend MESH_CONFIG command to access non-volatile configuration
Javier Cardona [Sat, 17 May 2008 07:55:10 +0000 (00:55 -0700)]
libertas: Extend MESH_CONFIG command to access non-volatile configuration

This patch is based on a patch from Shailendra Govardhan and Brian Cavagnolo.
It extends the MESH_CONFIG command to configure non-volatile parameters on
libertas devices that support them (e.g. OLPC Active Antenna).

This patch only implements the driver/firmware interface.

See http://dev.laptop.org/ticket/6823 for minimal testing results and known
issues.

Signed-off-by: Javier Cardona <javier@cozybit.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: fix multicast filtering on eth and msh interfaces
David Woodhouse [Tue, 20 May 2008 12:32:45 +0000 (13:32 +0100)]
libertas: fix multicast filtering on eth and msh interfaces

We weren't properly handling multicast on the mesh interface. Fix that,
which involves setting up the hardware to use the union of dev->mc_list
for both eth%d and msh%d devices.

This means we can't do it directly from ->set_multicast_list() because
we'd need to lock the other device to read its list, and we can't do
that because it might deadlock. So punt the actual work to keventd.

Also, invoke the same when taking an interface down; for some reason the
core calls ->set_multicast_list while IFF_UP is still set in dev->flags
when we're taking it down, so its addresses don't get removed then.

We also convert MAC_MULTICAST_ADR to a direct command while we're at it,
removing one more entry from the big switch statement in the deprecated
lbs_prepare_and_send_command() function.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agolibertas: Increase priority of 'unknown command' warnings
David Woodhouse [Tue, 20 May 2008 10:47:16 +0000 (11:47 +0100)]
libertas: Increase priority of 'unknown command' warnings

Using the deprecated lbs_prepare_and_send_command() function for a
command which it doesn't understand is an error; complain loudly about
it even when we're not debugging.

The mesh stats bug, where we converted MESH_ACCESS to a direct command
but accidentally missed one user which was still trying to do it through
lbs_prepare_and_send_command(), would have been caught a lot quicker if
we'd done this sooner. Such bugs aren't entirely unlikely in future too,
as we convert more code to stop using this function.

Signed-off-by: David Woodhouse <dwmw2@infradead.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Release rt2x00 2.1.6
Ivo van Doorn [Sat, 10 May 2008 11:45:50 +0000 (13:45 +0200)]
rt2x00: Release rt2x00 2.1.6

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Remove extra +
Ivo van Doorn [Sat, 10 May 2008 11:45:58 +0000 (13:45 +0200)]
rt2x00: Remove extra +

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Merge RX and TX entry private data
Ivo van Doorn [Sat, 10 May 2008 11:46:03 +0000 (13:46 +0200)]
rt2x00: Merge RX and TX entry private data

With the pending removal of the tx_control structure
we can merge the RX and TX entry private data structure
in advance. This will temporarily increase the required
memory for the queue, but that overhead will only be limited.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Remove redundant flags/dev_flags initializations
Ivo van Doorn [Sat, 10 May 2008 11:46:08 +0000 (13:46 +0200)]
rt2x00: Remove redundant flags/dev_flags initializations

the rxdesc structure is properly memsetted before passed to
the driver. This means we don't have to reinitialize the flags
and dev_flags fields in the drivers again.
This will prevent problems when the rxdone handler is adding
flags in a earlier status and will make the code look nicer
when we are adding more read attributes in the rxdone handler
in the driver.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Split rt2x00lib_write_tx_desc()
Ivo van Doorn [Sat, 10 May 2008 11:46:13 +0000 (13:46 +0200)]
rt2x00: Split rt2x00lib_write_tx_desc()

Split rt2x00lib_write_tx_desc() up into a TX descriptor initializor
and TX descriptor writer.

This split is required to properly allow mac80211 to move its
tx_control structure into the skb->cb array.
The rt2x00queue_create_tx_descriptor() function will read all tx control
information and convert it into a rt2x00 TX descriptor information structure.
After that function is complete, we have all information we needed from the
tx control structure and are free to start writing into the skb->cb array
for our own purposes.
rt2x00queue_write_tx_descriptor() will be in charge of really sending
the TX descriptor to the hardware and kicking the TX queue.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Only initialize the minimum needed fields of PCI TX descriptors.
Gertjan van Wingerde [Sat, 10 May 2008 11:44:14 +0000 (13:44 +0200)]
rt2x00: Only initialize the minimum needed fields of PCI TX descriptors.

In preparation of replacing the statically allocated data DMA buffers with DMA-mapped
skb's we need to change the TXD handling of the PCI drivers, by moving the programming
of the buffer address fields to the actual TXD writing at TX time, instead of at start-up
time.

Signed-off-by: Gertjan van Wingerde <gwingerde@kpnplanet.nl>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Preserve descriptor information after memmove()
Ivo van Doorn [Sat, 10 May 2008 11:43:38 +0000 (13:43 +0200)]
rt2x00: Preserve descriptor information after memmove()

Due to usage of memmove() in rt2x00usb the descriptor can become
corrupted because it is being overwritten by the data part.
Overall having the descriptor in front of the frame is a bad idea,
we can however use the skb->cb array for this task, since that
contains more then enough room to hold the entire descriptor and
preserve the information long enough.

After this we can also cleanup the alignment code a bit to make it
work a bit more flexible to allow for all kinds of odd header lengths.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Fix queue related oops in case of deselected mac80211 multi-queue feature.
Gertjan van Wingerde [Sat, 10 May 2008 11:43:33 +0000 (13:43 +0200)]
rt2x00: Fix queue related oops in case of deselected mac80211 multi-queue feature.

With the integration of the mac80211 multiqueue patches it has become possible that the
mac80211 layer modifies the number of TX queues that is stored inside the ieee80211_hw
structure, especially when multi-queue is not selected.

The rt2x00 drivers are not well suited to handle that situation, as they allocate the
queue structures before mac80211 has modified the number of queues it is going to use,
and also expect the number of allocated queues to match the hardware implementation.

Hence, ensure that rt2x00 maintains by itself the number of queues that the hardware
supports, and, at the same time, making is not dependent on the preservation of contents
inside a mac80211 structure.

Signed-off-by: Gertjan van Wingerde <gwingerde@kpnplanet.nl>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Remove ieee80211_tx_control argument from write_tx_desc()
Ivo van Doorn [Sat, 10 May 2008 11:42:31 +0000 (13:42 +0200)]
rt2x00: Remove ieee80211_tx_control argument from write_tx_desc()

Move the last remaining information details read from ieee80211_tx_control
in the drivers to the txentry_desc structure. After this we can
remove ieee80211_tx_control from the argument list for the callback function,
which makes it easier when the control information is moved into skb->cb

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: Fix TX status reporting
Ivo van Doorn [Sat, 10 May 2008 11:42:06 +0000 (13:42 +0200)]
rt2x00: Fix TX status reporting

The tx_status enumeration was broken since the introduction
of rt61pci. That driver uses different values to report the
status of the tx action.
This would lead to frames that were reported as success but
actually failed to be send out, or frames that were neither
successfull or failure which were reported as failure.

Fix this by change the TX status reporting and more explicitely
check for failure or success. Note that a third possibility is
added "unknown". Not all hardware (USB) can report the actual
TX status, for rt61pci some frames will receive this status
because the TXdone handler is never called for those frames.
This unknown will now be handled as neither success or failure,
so we no longer increment the failure counter while this conclusion
could not be determined from the real status of the frame.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agort2x00: trim skb_frame_desc to 32 bytes
Ivo van Doorn [Sat, 10 May 2008 11:41:32 +0000 (13:41 +0200)]
rt2x00: trim skb_frame_desc to 32 bytes

Remove frame_type from skb_frame_desc and pass it
as argument to rt2x00debug_dump_frame().

Change data_len and desc_len to unsigned short
to save another 4 bytes in skb_frame_desc. Note that
this was the only location where the data_len and
desc_len was not yet treated as unsigned short.

This trim is required to help mac80211 with adding
the TX control and TX status informtation into the
skb->cb structure. When that happens, drivers will
have approximately 40 bytes left to use freely.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agotipc: Cosmetic cleanup of topology service code
Allan Stephens [Mon, 19 May 2008 20:30:13 +0000 (13:30 -0700)]
tipc: Cosmetic cleanup of topology service code

This patch contains a set of cosmetic changes to TIPC's network
topology service subsystem, including:

- updates to comments (including copyright dates)
- re-ordering structure fields to group them more logically
- removal of optional debugging code that is no longer required
- minor changes to whitespace to conform to Linux coding conventions

Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agotipc: Consolidate subscriber & subscriber port references
Allan Stephens [Mon, 19 May 2008 20:29:47 +0000 (13:29 -0700)]
tipc: Consolidate subscriber & subscriber port references

This patch modifies TIPC's network topology service so that it
only requires a single reference table entry per subscriber
connection, rather than two.  This is achieved by letting the
reference to the server port communicating with the subscriber
act as the reference to the subscriber object itself.  (Since
the subscriber cannot exist without its port, and vice versa,
this dual role for the reference is perfectly natural.)  This
consolidation reduces the size of the reference table by 50%
in the default configuration.

Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agotipc: Fix bug in topology server byte swapping routine
Allan Stephens [Mon, 19 May 2008 20:29:06 +0000 (13:29 -0700)]
tipc: Fix bug in topology server byte swapping routine

This patch fixes TIPC's topology server so that it does byte swapping
correctly when endianness conversion is required.  (Note: This bug only
impacted an application if it issues a subscription request to a
topology server on another node, rather than the server on it's own
node; since the topology server is normally not accessible by off-node
applications, most TIPC applications were not impacted by the bug.)

Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agotipc: Add support for customized subscription endianness
Allan Stephens [Mon, 19 May 2008 20:28:32 +0000 (13:28 -0700)]
tipc: Add support for customized subscription endianness

This patch enables TIPC's topology server code to do customized
endianness conversions on a per-subscription basis.  (This
capability is needed to support the upcoming consolidation of
subscriber and subscription object references.)

Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agotipc: Add support for customized subscription overlap handling
Allan Stephens [Mon, 19 May 2008 20:27:31 +0000 (13:27 -0700)]
tipc: Add support for customized subscription overlap handling

This patch enables TIPC's topology server code to do customized
overlap detection handling on a per-subscription basis.  (This
capability is needed to support the upcoming introduction of
multi-cluster TIPC networks.)

Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2: Update version to 1.7.6.
Michael Chan [Sat, 17 May 2008 05:20:59 +0000 (22:20 -0700)]
bnx2: Update version to 1.7.6.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2: Pre-initialize struct cpu_reg.
Benjamin Li [Sat, 17 May 2008 05:20:27 +0000 (22:20 -0700)]
bnx2: Pre-initialize struct cpu_reg.

Instead of assigning values for the struct cpu_reg's at runtime,
we already know these values at compile time.  Therefore, we can use
designated initializers, to initialize these structures and not have
to incur this assignment cost at run-time.

Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2: Rename RX_COPY_THRESH to BNX2_RX_COPY_THRESH.
Benjamin Li [Sat, 17 May 2008 05:19:35 +0000 (22:19 -0700)]
bnx2: Rename RX_COPY_THRESH to BNX2_RX_COPY_THRESH.

To make the bnx2 code more consistent, all instances of
RX_COPY_THRESH have been changed to BNX2_RX_COPY_THRESH.

Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2: Remove the rx_offset field from the bnx2 structure.
Benjamin Li [Sat, 17 May 2008 05:18:57 +0000 (22:18 -0700)]
bnx2: Remove the rx_offset field from the bnx2 structure.

The rx_offset field is set to a constant value and initialized
only once.  By replacing all references to the rx_offset field,
we can eliminate rx_offset from the bnx2 structure.  This will
save 4 bytes for every bnx2 instance.

[Added parentheses to the definition of BNX2_RX_OFFSET, as noted
by Ben Hutchings.]

Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2: Add EEH PCI recovery.
Wendy Xiong [Sat, 17 May 2008 05:18:21 +0000 (22:18 -0700)]
bnx2: Add EEH PCI recovery.

Add PCI recovery functions to the driver.  The initial pci state is
also saved so the the MSI state can be restored during PCI recovery.

Signed-off-by: Wendy Xiong <wendyx@us.ibm.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2: Allow phy reset to be skipped during chip reset.
Michael Chan [Sat, 17 May 2008 05:17:45 +0000 (22:17 -0700)]
bnx2: Allow phy reset to be skipped during chip reset.

Andy Gospodarek <andy@greyhouse.net> found that netconsole would
panic when resetting bnx2 devices.

>From Andy:
"The issue is the bnx2_set_link in bnx2_init_nic will print a link-status
message before we are fully initialized and ready to start polling.
Polling is currently disabled in this state, but since the
__LINK_STATE_RX_SCHED is overloaded to not only try and disable polling
but also to make the system aware there is something waiting to be
polled, we really have to fix this in drivers.

The problematic call is the one to netif_rx_complete as it tries to
remove an entry from the poll_list when there isn't one."

While this netconsole problem should be fixed separately, we really
should not reset the PHY when changing ring sizes, MTU, or other
similar settings.  The PHY reset causes several seconds of unnecessary
link disruptions.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Acked-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Thu, 15 May 2008 07:52:37 +0000 (00:52 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next-2.6

16 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Thu, 15 May 2008 07:34:44 +0000 (00:34 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/davem/net-2.6

Conflicts:

drivers/net/wireless/iwlwifi/iwl-4965-rs.c
drivers/net/wireless/rt2x00/rt61pci.c

16 years agoisdn/capi: Return proper errnos on module init.
Marcin Slusarz [Thu, 15 May 2008 06:30:06 +0000 (23:30 -0700)]
isdn/capi: Return proper errnos on module init.

cdebug_init() is called from kcapi_init() which is module
initialization function, so it must return negative values on errors.

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Acked-by: Karsten Keil <kkeil@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>