linux-2.6
16 years agoAth5k: add AP mode
Jiri Slaby [Sun, 12 Oct 2008 20:54:10 +0000 (22:54 +0200)] 
Ath5k: add AP mode

Add support for AP mode. This involves:
- enablement in ath5k_beacon_config -- initialize beacon timer
- add AP to the supported modes in ath5k_add_interface
- handle beacon change even for AP in ath5k_config_interface
- remove useless test for IBSS in ath5k_beacon_update

Note that it doesn't enable the AP mode for the driver. It must
be enabled by NL80211_IFTYPE_AP bit added to interface_modes.

v2:
Fixed opmode constant (IEEE80211_ to NL80211_)

Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Cc: Nick Kossifidis <mickflemm@gmail.com>
Cc: Luis R. Rodriguez <mcgrof@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agop54/rtl8187: fix up the seqno patch
Johannes Berg [Sat, 11 Oct 2008 22:39:55 +0000 (00:39 +0200)] 
p54/rtl8187: fix up the seqno patch

Sorry about that, for some reason I didn't notice that I'd
left some unused variables in there.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: rewrite HT handling
Johannes Berg [Tue, 14 Oct 2008 14:58:37 +0000 (16:58 +0200)] 
mac80211: rewrite HT handling

The HT handling has the following deficiencies, which I've
(partially) fixed:
 * it always uses the AP info even if there is no AP,
   hence has no chance of working as an AP
 * it pretends to be HW config, but really is per-BSS
 * channel sanity checking is left to the drivers
 * it generally lets the driver control too much

HT enabling is still wrong with this patch if you have more than
one virtual STA mode interface, but that never happens currently.
Once WDS, IBSS or AP/VLAN gets HT capabilities, it will also be
wrong, see the comment in ieee80211_enable_ht().

Additionally, this fixes a number of bugs:
 * mac80211: ieee80211_set_disassoc doesn't notify the driver any
             more since the refactoring
 * iwl-agn-rs: always uses the HT capabilities from the wrong stuff
               mac80211 gives it rather than the actual peer STA
 * ath9k: a number of bugs resulting from the broken HT API

I'm not entirely happy with putting the HT capabilities into
struct ieee80211_sta as restricted to our own HT TX capabilities,
but I see no cleaner solution for now.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: move bss_conf into vif
Johannes Berg [Fri, 10 Oct 2008 23:51:51 +0000 (01:51 +0200)] 
mac80211: move bss_conf into vif

Move bss_conf into the vif struct so that drivers can
access it during ->tx without having to store it in
the private data or similar. No driver updates because
this is only for when they want to start using it.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: make retry limits part of hw config
Johannes Berg [Tue, 14 Oct 2008 17:17:54 +0000 (19:17 +0200)] 
mac80211: make retry limits part of hw config

Instead of having a separate callback, use the HW config callback
with a new flag to change retry limits.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwl3945: fix station stuff in RC algorithm
Johannes Berg [Fri, 10 Oct 2008 23:46:25 +0000 (01:46 +0200)] 
iwl3945: fix station stuff in RC algorithm

Probably bugs I added.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agonl80211: export HT capabilities
Johannes Berg [Thu, 9 Oct 2008 10:20:13 +0000 (12:20 +0200)] 
nl80211: export HT capabilities

This exports the local HT capabilities in nl80211.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: provide sequence numbers
Johannes Berg [Fri, 10 Oct 2008 11:21:59 +0000 (13:21 +0200)] 
mac80211: provide sequence numbers

I've come to think that not providing sequence numbers for
the normal STA mode case was a mistake, at least two drivers
now had to implement code they wouldn't otherwise need, and
I believe at76_usb and adm8211 might be broken.

This patch makes mac80211 assign a sequence number to all
those frames that need one except beacons. That means that
if a driver only implements modes that do not do beaconing
it need not worry about the sequence number.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agop54: Move rx_mtu to struct bootrec_desc
Larry Finger [Fri, 10 Oct 2008 00:38:52 +0000 (17:38 -0700)] 
p54: Move rx_mtu to struct bootrec_desc

The patch entitled "[PATCH] p54: Fix sparse warnings" added the __le16
variable rx_mtu to struct bootrec, but it could equally well be placed
in the struct bootrec_desc, which overlays the 'data' section of bootrec.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agorfkill: rate-limit rfkill-input workqueue usage (v3)
Henrique de Moraes Holschuh [Thu, 9 Oct 2008 21:15:33 +0000 (18:15 -0300)] 
rfkill: rate-limit rfkill-input workqueue usage (v3)

Limit the number of "expensive" rfkill workqueue operations per second, in
order to not hog system resources too much when faced with a rogue source
of rfkill input events.

The old rfkill-input code (before it was refactored) had such a limit in
place.  It used to drop new events that were past the rate limit.  This
behaviour was not implemented as an anti-DoS measure, but rather as an
attempt to work around deficiencies in input device drivers which would
issue multiple KEY_FOO events too soon for a given key FOO (i.e. ones that
do not implement mechanical debouncing properly).

However, we can't really expect such issues to be worked around by every
input handler out there, and also by every userspace client of input
devices.  It is the input device driver's responsability to do debouncing
instead of spamming the input layer with bogus events.

The new limiter code is focused only on anti-DoS behaviour, and tries to
not lose events (instead, it coalesces them when possible).

The transmitters are updated once every 200ms, maximum.  Care is taken not
to delay a request to _enter_ rfkill transmitter Emergency Power Off (EPO)
mode.

If mistriggered (e.g. by a jiffies counter wrap), the code delays processing
*once* by 200ms.

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Ivo van Doorn <IvDoorn@gmail.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agorfkill: honour EPO state when resuming a rfkill controller
Henrique de Moraes Holschuh [Thu, 9 Oct 2008 21:15:32 +0000 (18:15 -0300)] 
rfkill: honour EPO state when resuming a rfkill controller

rfkill_resume() would always restore the rfkill controller state to its
pre-suspend state.

Now that we know when we are under EPO, kick the rfkill controller to
SOFT_BLOCKED state instead of to its pre-suspend state when it is resumed
while EPO mode is active.

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agorfkill: add master_switch_mode and EPO lock to rfkill and rfkill-input
Henrique de Moraes Holschuh [Fri, 10 Oct 2008 00:49:33 +0000 (21:49 -0300)] 
rfkill: add master_switch_mode and EPO lock to rfkill and rfkill-input

Add of software-based sanity to rfkill and rfkill-input so that it can
reproduce what hardware-based EPO switches do, blocking all transmitters
and locking down any further attempts to unblock them until the switch is
deactivated.

rfkill-input is responsible for issuing the EPO control requests, like
before.

While an rfkill EPO is active, all transmitters are locked to one of the
BLOCKED states and all attempts to change that through the rfkill API
(userspace and kernel) will be either ignored or return -EPERM errors.

The lock will be released upon receipt of EV_SW SW_RFKILL_ALL ON by
rfkill-input, or should modular rfkill-input be unloaded.

This makes rfkill and rfkill-input extend the operation of an existing
wireless master kill switch to all wireless devices in the system, even
those that are not under hardware or firmware control.

Since the above is the expected operational behavior for the master rfkill
switch, the EPO lock functionality is not optional.

Also, extend rfkill-input to allow for three different behaviors when it
receives an EV_SW SW_RFKILL_ALL ON input event.  The user can set which
behavior he wants through the master_switch_mode parameter:

master_switch_mode = 0: EV_SW SW_RFKILL_ALL ON just unlocks rfkill
controller state changes (so that the rfkill userspace and kernel APIs can
now be used to change rfkill controller states again), but doesn't change
any of their states (so they will all remain blocked).  This is the safest
mode of operation, as it requires explicit operator action to re-enable a
transmitter.

master_switch_mode = 1: EV_SW SW_RFKILL_ALL ON causes rfkill-input to
attempt to restore the system to the state before the last EV_SW
SW_RFKILL_ALL OFF event, or to the default global states if no EV_SW
SW_RFKILL_ALL OFF ever happened.   This is the recommended mode of
operation for laptops.

master_switch_mode = 2: tries to unblock all rfkill controllers (i.e.
enable all transmitters) when an EV_SW SW_RFKILL_ALL ON event is received.
This is the default mode of operation, as it mimics the previous behavior
of rfkill-input.

In order to implement these features in a clean way, the entire event
handling of rfkill-input was refactored into a single worker function.

Protection against input event DoS (repeatedly firing rfkill events for
rfkill-input to process) was removed during the code refactoring.  It will
be added back in a future patch.

Note that with these changes, rfkill-input doesn't need to explicitly
handle any radio types for which KEY_<radio type> or SW_<radio type> events
do not exist yet.

Code to handle EV_SW SW_{WLAN,WWAN,BLUETOOTH,WIMAX,...} was added as it
might be needed in the future (and its implementation is not that obvious),
but is currently #ifdef'd out to avoid wasting resources.

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Ivo van Doorn <IvDoorn@gmail.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agorfkill: export global states to rfkill-input
Henrique de Moraes Holschuh [Thu, 9 Oct 2008 21:15:30 +0000 (18:15 -0300)] 
rfkill: export global states to rfkill-input

Export the the global switch states to rfkill-input.  This is needed to
properly implement KEY_* handling without disregarding the initial state.

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agorfkill: use killable locks instead of interruptible
Henrique de Moraes Holschuh [Thu, 9 Oct 2008 21:15:29 +0000 (18:15 -0300)] 
rfkill: use killable locks instead of interruptible

Apparently, many applications don't expect to get EAGAIN from fd read/write
operations, since POSIX doesn't mandate it.

Use mutex_lock_killable instead of mutex_lock_interruptible, which won't
cause issues.

Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br>
Cc: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: introduce hw config change flags
Johannes Berg [Thu, 9 Oct 2008 10:18:51 +0000 (12:18 +0200)] 
mac80211: introduce hw config change flags

This makes mac80211 notify the driver which configuration
actually changed, e.g. channel etc.

No driver changes, this is just plumbing, driver authors are
expected to act on this if they want to.

Also remove the HW CONFIG debug printk, it's incorrect, often
we configure something else.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: kill hw.conf.antenna_sel_{rx,tx}
Johannes Berg [Thu, 9 Oct 2008 10:18:04 +0000 (12:18 +0200)] 
mac80211: kill hw.conf.antenna_sel_{rx,tx}

Never actually used.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years ago802.11: clean up/fix HT support
Johannes Berg [Thu, 9 Oct 2008 10:13:49 +0000 (12:13 +0200)] 
802.11: clean up/fix HT support

This patch cleans up a number of things:
 * the unusable definition of the HT capabilities/HT information
   information elements
 * variable names that are hard to understand
 * mac80211: move ieee80211_handle_ht to ht.c and remove the unused
             enable_ht parameter
 * mac80211: fix bug with MCS rate 32 in ieee80211_handle_ht
 * mac80211: fix bug with casting the result of ieee80211_bss_get_ie
             to an information element _contents_ rather than the
             whole element, add size checking (another out-of-bounds
             access bug fixed!)
 * mac80211: remove some unused return values in favour of BUG_ON
             checking
 * a few minor other things

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agop54: honour bss_info_changed's short slot time settings
Christian Lamparter [Wed, 8 Oct 2008 18:52:22 +0000 (20:52 +0200)] 
p54: honour bss_info_changed's short slot time settings

This patch was made on behalf of Johannes request.
"mac80211 and IEEE80211_CONF_SHORT_SLOT_TIME"

Of course, bss_info_changed provides some more useful data.
e.g.: basic_rates, dtim_period, beacon_int and maybe even more.
Everything can be hooked up if it's necessary.

Signed-off-by: Christian Lamparter <chunkeey@web.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: use HWSIM_STA_MAGIC in hwsim_check_sta_magic() and hwsim_set_sta_magic().
Rami Rosen [Wed, 8 Oct 2008 09:18:27 +0000 (11:18 +0200)] 
mac80211: use HWSIM_STA_MAGIC in hwsim_check_sta_magic() and hwsim_set_sta_magic().

This patch fixes a typo in mac80211_hwsim.c:
use HWSIM_STA_MAGIC in hwsim_check_sta_magic() and hwsim_set_sta_magic()
instead of HWSIM_VIF_MAGIC.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agoiwlwifi: refactor rxon time command
Tomas Winkler [Wed, 8 Oct 2008 01:37:30 +0000 (09:37 +0800)] 
iwlwifi: refactor rxon time command

This patch refactors rxon time command. It removes the usage of union tsf
in favor of u64 value and hopefully makes code more readable.  There are
no functional changes in this patch.

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 agoiwlwifi: take a valid antenna upon rate scale init
Tomas Winkler [Wed, 8 Oct 2008 01:37:29 +0000 (09:37 +0800)] 
iwlwifi: take a valid antenna upon rate scale init

This patch selects a valid antennae upon rate scale init. This solves a
SYSASSERT complaining that the driver is setting a non valid antenna in
the LQ CMD.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
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 agoiwlwifi: make initial calibration set configurable
Tomas Winkler [Wed, 8 Oct 2008 01:37:27 +0000 (09:37 +0800)] 
iwlwifi: make initial calibration set configurable

This patch adds ability to configure initial calibration set. Not all HW
supported by iwlwifi use the same calibration set, XTAL is one example.
Some clean ups are also included in this patch.

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 agomac80211: remove unused declaration of struct sta_attribute.
Rami Rosen [Tue, 7 Oct 2008 14:29:11 +0000 (16:29 +0200)] 
mac80211: remove unused declaration of struct sta_attribute.

This patch removes unused definition of struct sta_attribute
in net/mac80211/ieee80211_i.h.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: fix short slot handling
Johannes Berg [Wed, 8 Oct 2008 08:59:33 +0000 (10:59 +0200)] 
mac80211: fix short slot handling

This patch makes mac80211 handle short slot requests from the AP
properly. Also warn about uses of IEEE80211_CONF_SHORT_SLOT_TIME
and optimise out the code since it cannot ever be hit anyway.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: remove max_antenna_gain config
Johannes Berg [Tue, 7 Oct 2008 10:04:35 +0000 (12:04 +0200)] 
mac80211: remove max_antenna_gain config

The antenna gain isn't exactly configurable, despite the belief of
some unnamed individual who thinks that the EEPROM might influence
it.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: clean up ieee80211_hw_config errors
Johannes Berg [Tue, 7 Oct 2008 10:04:34 +0000 (12:04 +0200)] 
mac80211: clean up ieee80211_hw_config errors

Warn when ieee80211_hw_config returns an error, it shouldn't
happen; remove a number of printks that would happen in such
a case and one printk that is user-triggerable.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: remove wiphy_to_hw
Johannes Berg [Tue, 7 Oct 2008 10:04:33 +0000 (12:04 +0200)] 
mac80211: remove wiphy_to_hw

This isn't used by anyone, if we ever need it we can add
it back, until then it's useless.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: minor code cleanups
Johannes Berg [Tue, 7 Oct 2008 10:04:32 +0000 (12:04 +0200)] 
mac80211: minor code cleanups

Nothing very interesting, some checkpatch inspired stuff,
some other things.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: remove writable debugs mesh parameters
Johannes Berg [Tue, 7 Oct 2008 10:04:31 +0000 (12:04 +0200)] 
mac80211: remove writable debugs mesh parameters

These parameters shouldn't be configurable via debugfs, if they
need to be configurable nl80211 support has to be added, if not
then they don't need to be writable here either.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: Javier Cardona <javier@cozybit.com>
Cc: Luis Carlos Cobo <luisca@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomac80211: remove aggregation status write support from debugfs
Johannes Berg [Tue, 7 Oct 2008 10:04:30 +0000 (12:04 +0200)] 
mac80211: remove aggregation status write support from debugfs

This code uses static variables and thus cannot be kept.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
16 years agomisc: replace NIPQUAD()
Harvey Harrison [Fri, 31 Oct 2008 07:56:49 +0000 (00:56 -0700)] 
misc: replace NIPQUAD()

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agofs: replace NIPQUAD()
Harvey Harrison [Fri, 31 Oct 2008 07:56:28 +0000 (00:56 -0700)] 
fs: replace NIPQUAD()

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agodrivers: replace NIPQUAD()
Harvey Harrison [Fri, 31 Oct 2008 07:56:00 +0000 (00:56 -0700)] 
drivers: replace NIPQUAD()

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonet: replace NIPQUAD() in net/*/
Harvey Harrison [Fri, 31 Oct 2008 07:54:56 +0000 (00:54 -0700)] 
net: replace NIPQUAD() in net/*/

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonet: replace NIPQUAD() in net/netfilter/
Harvey Harrison [Fri, 31 Oct 2008 07:54:29 +0000 (00:54 -0700)] 
net: replace NIPQUAD() in net/netfilter/

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonet: replace NIPQUAD() in net/ipv4/ net/ipv6/
Harvey Harrison [Fri, 31 Oct 2008 07:53:57 +0000 (00:53 -0700)] 
net: replace NIPQUAD() in net/ipv4/ net/ipv6/

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonet: replace NIPQUAD() in net/ipv4/netfilter/
Harvey Harrison [Fri, 31 Oct 2008 07:53:08 +0000 (00:53 -0700)] 
net: replace NIPQUAD() in net/ipv4/netfilter/

Using NIPQUAD() with NIPQUAD_FMT, %d.%d.%d.%d or %u.%u.%u.%u
can be replaced with %pI4

Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Add peek emulation for non-work-conserving qdiscs.
Jarek Poplawski [Fri, 31 Oct 2008 07:47:01 +0000 (00:47 -0700)] 
pkt_sched: Add peek emulation for non-work-conserving qdiscs.

This patch adds qdisc_peek_dequeued() wrapper to emulate peek method
with qdisc->dequeue() and storing "peeked" skb in qdisc->gso_skb until
dequeuing. This is mainly for compatibility reasons not to break some
strange configs because peeking is expected for non-work-conserving
parent qdiscs to query work-conserving child qdiscs.

This implementation requires using qdisc_dequeue_peeked() wrapper
instead of directly calling qdisc->dequeue() for all qdiscs ever
querried with qdisc->ops->peek() or qdisc_peek_dequeued().

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Use qdisc->ops->peek() instead of ->dequeue() & ->requeue()
Jarek Poplawski [Fri, 31 Oct 2008 07:46:19 +0000 (00:46 -0700)] 
pkt_sched: Use qdisc->ops->peek() instead of ->dequeue() & ->requeue()

Use qdisc->ops->peek() instead of ->dequeue() & ->requeue() pair.
After this patch the only remaining user of qdisc->ops->requeue() is
netem_enqueue(). Based on ideas of Herbert Xu, Patrick McHardy and
David S. Miller.

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Add qdisc->ops->peek() implementation.
Jarek Poplawski [Fri, 31 Oct 2008 07:45:55 +0000 (00:45 -0700)] 
pkt_sched: Add qdisc->ops->peek() implementation.

Add qdisc->ops->peek() implementation for work-conserving qdiscs.
With feedback from Patrick McHardy.

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: sch_generic: Add generic qdisc->ops->peek() implementation.
Jarek Poplawski [Fri, 31 Oct 2008 07:45:27 +0000 (00:45 -0700)] 
pkt_sched: sch_generic: Add generic qdisc->ops->peek() implementation.

With feedback from Patrick McHardy.

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: Add ->peek() methods for fifo, prio and SFQ qdiscs.
Patrick McHardy [Fri, 31 Oct 2008 07:44:18 +0000 (00:44 -0700)] 
pkt_sched: Add ->peek() methods for fifo, prio and SFQ qdiscs.

From: Patrick McHardy <kaber@trash.net>

Just as a demonstration how easy adding a peek operation to the
work-conserving qdiscs actually is. It doesn't need to keep or change
any internal state in many cases thanks to the guarantee that the
packet will either be dequeued or, if another packet arrives, the
upper qdisc will immediately ->peek again to reevaluate the state.

(This is only slightly modified Patrick's patch.)

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agopkt_sched: sch_generic: Add Qdisc_ops peek() method.
Jarek Poplawski [Fri, 31 Oct 2008 07:43:45 +0000 (00:43 -0700)] 
pkt_sched: sch_generic: Add Qdisc_ops peek() method.

Add Qdisc_ops peek() method in order to replace requeuing.

Based on ideas and patches of Herbert Xu, Patrick McHardy and
David S. Miller.

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoxfrm: remove unused struct xfrm_policy::next
Alexey Dobriyan [Fri, 31 Oct 2008 07:42:25 +0000 (00:42 -0700)] 
xfrm: remove unused struct xfrm_policy::next

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoxfrm: C99 for xfrm_dev_notifier
Alexey Dobriyan [Fri, 31 Oct 2008 07:41:59 +0000 (00:41 -0700)] 
xfrm: C99 for xfrm_dev_notifier

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Fri, 31 Oct 2008 07:17:34 +0000 (00:17 -0700)] 
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

Conflicts:

drivers/net/wireless/p54/p54common.c

16 years agoxfrm: do not leak ESRCH to user space
fernando@oss.ntt.co [Thu, 23 Oct 2008 04:27:19 +0000 (04:27 +0000)] 
xfrm: do not leak ESRCH to user space

I noticed that, under certain conditions, ESRCH can be leaked from the
xfrm layer to user space through sys_connect. In particular, this seems
to happen reliably when the kernel fails to resolve a template either
because the AF_KEY receive buffer being used by racoon is full or
because the SA entry we are trying to use is in XFRM_STATE_EXPIRED
state.

However, since this could be a transient issue it could be argued that
EAGAIN would be more appropriate. Besides this error code is not even
documented in the man page for sys_connect (as of man-pages 3.07).

Signed-off-by: Fernando Luis Vazquez Cao <fernando@oss.ntt.co.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonet: Really remove all of LOOPBACK_TSO code.
David S. Miller [Fri, 31 Oct 2008 07:00:33 +0000 (00:00 -0700)] 
net: Really remove all of LOOPBACK_TSO code.

As noticed by Saikiran Madugula, commit 7447ef63cf2dfdc444f4c72ae13f604350b2e25f
("loopback: Remove rest of LOOPBACK_TSO code.") got rid of
emulate_large_send_offload() but didn't get rid of the call
site as well.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'master' of git://git.infradead.org/users/pcmoore/lblnet-2.6
David S. Miller [Fri, 31 Oct 2008 06:57:40 +0000 (23:57 -0700)] 
Merge branch 'master' of git://git.infradead.org/users/pcmoore/lblnet-2.6

16 years agonetfilter: nf_conntrack_proto_gre: switch to register_pernet_gen_subsys()
Alexey Dobriyan [Fri, 31 Oct 2008 06:55:44 +0000 (23:55 -0700)] 
netfilter: nf_conntrack_proto_gre: switch to register_pernet_gen_subsys()

register_pernet_gen_device() can't be used is nf_conntrack_pptp module is
also used (compiled in or loaded).

Right now, proto_gre_net_exit() is called before nf_conntrack_pptp_net_exit().
The former shutdowns and frees GRE piece of netns, however the latter
absolutely needs it to flush keymap. Oops is inevitable.

Switch to shiny new register_pernet_gen_subsys() to get correct ordering in
netns ops list.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonetns: add register_pernet_gen_subsys/unregister_pernet_gen_subsys
Alexey Dobriyan [Fri, 31 Oct 2008 06:55:16 +0000 (23:55 -0700)] 
netns: add register_pernet_gen_subsys/unregister_pernet_gen_subsys

netns ops which are registered with register_pernet_gen_device() are
shutdown strictly before those which are registered with
register_pernet_subsys(). Sometimes this leads to opposite (read: buggy)
shutdown ordering between two modules.

Add register_pernet_gen_subsys()/unregister_pernet_gen_subsys() for modules
which aren't elite enough for entry in struct net, and which can't use
register_pernet_gen_device(). PPTP conntracking module is such one.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agonet: delete excess kernel-doc notation
Randy Dunlap [Fri, 31 Oct 2008 06:54:35 +0000 (23:54 -0700)] 
net: delete excess kernel-doc notation

Remove excess kernel-doc function parameters from networking header
& driver files:

Warning(include/net/sock.h:946): Excess function parameter or struct member 'sk' description in 'sk_filter_release'
Warning(include/linux/netdevice.h:1545): Excess function parameter or struct member 'cpu' description in 'netif_tx_lock'
Warning(drivers/net/wan/z85230.c:712): Excess function parameter or struct member 'regs' description in 'z8530_interrupt'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge branch 'davem-fixes' of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik...
David S. Miller [Fri, 31 Oct 2008 06:50:18 +0000 (23:50 -0700)] 
Merge branch 'davem-fixes' of /linux/kernel/git/jgarzik/netdev-2.6

16 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Fri, 31 Oct 2008 06:35:23 +0000 (23:35 -0700)] 
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

16 years agopppoe: Fix socket leak.
David S. Miller [Fri, 31 Oct 2008 06:11:44 +0000 (23:11 -0700)] 
pppoe: Fix socket leak.

Move SKB trim before we lookup the socket so we don't have to
put it on failure.

Based upon an initial patch by Jarek Poplawski and suggestions
from Herbert Xu.

Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agogianfar: Don't reset TBI<->SerDes link if it's already up
Trent Piepho [Fri, 31 Oct 2008 01:17:07 +0000 (18:17 -0700)] 
gianfar: Don't reset TBI<->SerDes link if it's already up

The link may be up already via the chip's reset strapping, or though action
of U-Boot, or from the last time the interface was brought up.  Resetting
the link causes it to go down for several seconds.  This can significantly
increase the time from power-on to DHCP completion and a device being
accessible to the network.

Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Acked-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agogianfar: Fix race in TBI/SerDes configuration
Trent Piepho [Fri, 31 Oct 2008 01:17:06 +0000 (18:17 -0700)] 
gianfar: Fix race in TBI/SerDes configuration

The init_phy() function attaches to the PHY, then configures the
SerDes<->TBI link (in SGMII mode).  The TBI is on the MDIO bus with the PHY
(sort of) and is accessed via the gianfar's MDIO registers, using the
functions gfar_local_mdio_read/write(), which don't do any locking.

The previously attached PHY will start a work-queue on a timer, and
probably an irq handler as well, which will talk to the PHY and thus use
the MDIO bus.  This uses phy_read/write(), which have locking, but not
against the gfar_local_mdio versions.

The result is that PHY code will try to use the MDIO bus at the same time
as the SerDes setup code, corrupting the transfers.

Setting up the SerDes before attaching to the PHY will insure that there is
no race between the SerDes code and *our* PHY, but doesn't fix everything.
Typically the PHYs for all gianfar devices are on the same MDIO bus, which
is associated with the first gianfar device.  This means that the first
gianfar's SerDes code could corrupt the MDIO transfers for a different
gianfar's PHY.

The lock used by phy_read/write() is contained in the mii_bus structure,
which is pointed to by the PHY.  This is difficult to access from the
gianfar drivers, as there is no link between a gianfar device and the
mii_bus which shares the same MDIO registers.  As far as the device layer
and drivers are concerned they are two unrelated devices (which happen to
share registers).

Generally all gianfar devices' PHYs will be on the bus associated with the
first gianfar.  But this might not be the case, so simply locking the
gianfar's PHY's mii bus might not lock the mii bus that the SerDes setup
code is going to use.

We solve this by having the code that creates the gianfar platform device
look in the device tree for an mdio device that shares the gianfar's
registers.  If one is found the ID of its platform device is saved in the
gianfar's platform data.

A new function in the gianfar mii code, gfar_get_miibus(), can use the bus
ID to search through the platform devices for a gianfar_mdio device with
the right ID.  The platform device's driver data is the mii_bus structure,
which the SerDes setup code can use to lock the current bus.

Signed-off-by: Trent Piepho <tpiepho@freescale.com>
CC: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agopci: use pci_ioremap_bar() in drivers/net
Arjan van de Ven [Tue, 21 Oct 2008 04:42:39 +0000 (21:42 -0700)] 
pci: use pci_ioremap_bar() in drivers/net

Use the newly introduced pci_ioremap_bar() function in drivers/net.
pci_ioremap_bar() just takes a pci device and a bar number, with the goal
of making it really hard to get wrong, while also having a central place
to stick sanity checks.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agotc35815: Define more Rx status bits
Atsushi Nemoto [Tue, 28 Oct 2008 13:30:23 +0000 (22:30 +0900)] 
tc35815: Define more Rx status bits

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agocxgb3 - enable lro control through ethtool
Divy Le Ray [Wed, 29 Oct 2008 05:40:32 +0000 (22:40 -0700)] 
cxgb3 - enable lro control through ethtool

Implement ethtool's get_flags and set_flags methods.
It enables ethtool to control the LRO settings.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoe100: adapt to the reworked PCI PM
Rafael J. Wysocki [Wed, 29 Oct 2008 21:22:18 +0000 (14:22 -0700)] 
e100: adapt to the reworked PCI PM

Adapt the e100 driver to the reworked PCI PM

* Use the observation that it is sufficient to call pci_enable_wake()
  once, unless it fails

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Tested-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Frans Pop <elendil@planet.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoThe overdue eepro100 removal.
Adrian Bunk [Wed, 29 Oct 2008 21:22:15 +0000 (14:22 -0700)] 
The overdue eepro100 removal.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoskge: adapt skge to use reworked PCI PM
Rafael J. Wysocki [Wed, 29 Oct 2008 21:22:14 +0000 (14:22 -0700)] 
skge: adapt skge to use reworked PCI PM

Adapt the skge driver to the reworked PCI PM

* Use device_set_wakeup_enable() and friends as needed
* Remove an open-coded reference to the standard PCI PM registers
* Use pci_prepare_to_sleep() and pci_back_from_sleep() in the
  ->suspend() and ->resume() callbacks
* Use the observation that it is sufficient to call pci_enable_wake()
  once, unless it fails

Tested on Asus L5D (Yukon-Lite rev 7).

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Stephen Hemminger <shemminger@linux-foundation.org>
Cc: Jeff Garzik <jeff@garzik.org>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoat91_ether: request/free GPIO for PHY interrupt
David Brownell [Mon, 27 Oct 2008 21:11:34 +0000 (14:11 -0700)] 
at91_ether: request/free GPIO for PHY interrupt

When the at91_ether driver is using a GPIO for its PHY interrupt,
be sure to request (and later, if needed, free) that GPIO.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoamd8111e: fix dma_free_coherent context
Chunbo Luo [Tue, 28 Oct 2008 01:51:46 +0000 (09:51 +0800)] 
amd8111e: fix dma_free_coherent context

Acoording commit aa24886e379d2b641c5117e178b15ce1d5d366ba,
dma_free_coherent() need irqs enabled.

This patch fix following warning messages:

WARNING: at linux/arch/x86/kernel/pci-dma.c:376 dma_free_coherent+0xaa/0xb0()

Call Trace:
 [<ffffffff8023f80f>] warn_on_slowpath+0x5f/0x90
 [<ffffffff80496ffa>] ? __kfree_skb+0x3a/0xa0
 [<ffffffff802a4723>] ? discard_slab+0x23/0x40
 [<ffffffff8021274a>] dma_free_coherent+0xaa/0xb0
 [<ffffffff8043668f>] amd8111e_close+0x10f/0x1b0
 [<ffffffff8049f3ae>] dev_close+0x5e/0xb0
 [<ffffffff8049efa1>] dev_change_flags+0xa1/0x1e0
 [<ffffffff806b2171>] ic_close_devs+0x36/0x4e
 [<ffffffff806b29ee>] ip_auto_config+0x581/0x10f3
 [<ffffffff803a6e19>] ? kobject_add+0x69/0x90
 [<ffffffff803a698a>] ? kobject_get+0x1a/0x30
 [<ffffffff803a785b>] ? kobject_uevent+0xb/0x10
 [<ffffffff803a6c62>] ? kset_register+0x52/0x60
 [<ffffffff803a6f9b>] ? kset_create_and_add+0x6b/0xa0
 [<ffffffff804e2e74>] ? tcp_ca_find+0x24/0x50
 [<ffffffff806b246d>] ? ip_auto_config+0x0/0x10f3
 [<ffffffff8020903c>] _stext+0x3c/0x150
 [<ffffffff802772d3>] ? register_irq_proc+0xd3/0xf0
 [<ffffffff802f0000>] ? mb_cache_create+0x80/0x1f0
 [<ffffffff80688693>] kernel_init+0x141/0x1b8
 [<ffffffff80688552>] ? kernel_init+0x0/0x1b8
 [<ffffffff8020d609>] child_rip+0xa/0x11
 [<ffffffff80688552>] ? kernel_init+0x0/0x1b8
 [<ffffffff80688552>] ? kernel_init+0x0/0x1b8
 [<ffffffff8020d5ff>] ? child_rip+0x0/0x11

Signed-off-by: Chunbo Luo <chunbo.luo@windriver.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoatl1: fix vlan tag regression
Jay Cliburn [Wed, 29 Oct 2008 16:01:36 +0000 (11:01 -0500)] 
atl1: fix vlan tag regression

Commit 401c0aabec4b97320f962a0161a846d230a6f7aa introduced a regression
in the atl1 driver by storing the VLAN tag in the wrong TX descriptor
field.

This patch causes the VLAN tag to be stored in its proper location.

Tested-by: Ramon Casellas <ramon.casellas@cttc.es>
Signed-off-by: Jay Cliburn <jacliburn@bellsouth.net>
Cc: stable@kernel.org
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoSMC91x: delete unused local variable "lp"
Mike Frysinger [Wed, 29 Oct 2008 13:56:06 +0000 (09:56 -0400)] 
SMC91x: delete unused local variable "lp"

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agomyri10ge: fix stop/go mmio ordering
Brice Goglin [Thu, 30 Oct 2008 07:59:33 +0000 (08:59 +0100)] 
myri10ge: fix stop/go mmio ordering

Use mmiowb() to ensure "stop" and "go" commands are sent in order on ia64.

Signed-off-by: Brice Goglin <brice@myri.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agobonding: fix panic when taking bond interface down before removing module
Andy Gospodarek [Fri, 31 Oct 2008 00:41:16 +0000 (17:41 -0700)] 
bonding: fix panic when taking bond interface down before removing module

A panic was discovered with bonding when using mode 5 or 6 and trying to
remove the slaves from the bond after the interface was taken down.
When calling 'ifconfig bond0 down' the following happens:

    bond_close()
        bond_alb_deinitialize()
            tlb_deinitialize()
kfree(bond_info->tx_hashtbl)
                bond_info->tx_hashtbl = NULL

Unfortunately if there are still slaves in the bond, when removing the
module the following happens:

    bonding_exit()
        bond_free_all()
            bond_release_all()
                bond_alb_deinit_slave()
                    tlb_clear_slave()
                        tx_hash_table = BOND_ALB_INFO(bond).tx_hashtbl
u32 next_index = tx_hash_table[index].next

As you might guess we panic when trying to access a few entries into the
table that no longer exists.

I experimented with several options (like moving the calls to
tlb_deinitialize somewhere else), but it really makes the most sense to
be part of the bond_close routine.  It also didn't seem logical move
tlb_clear_slave around too much, so the simplest option seems to add a
check in tlb_clear_slave to make sure we haven't already wiped the
tx_hashtbl away before searching for all the non-existent hash-table
entries that used to point to the slave as the output interface.

Signed-off-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agobonding: Clean up resource leaks
Jay Vosburgh [Fri, 31 Oct 2008 00:41:15 +0000 (17:41 -0700)] 
bonding: Clean up resource leaks

This patch reworks the resource free logic performed at the time
a bonding device is released.  This (a) closes two resource leaks, one
for workqueues and one for multicast lists, and (b) improves commonality
of code between the "destroy one" and "destroy all" paths by performing
final free activity via destructor instead of explicitly (and differently)
in each path.

"Sean E. Millichamp" <sean@bruenor.org> reported the workqueue
leak, and included a different patch.

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agobonding: fix miimon failure counter
Jay Vosburgh [Fri, 31 Oct 2008 00:41:14 +0000 (17:41 -0700)] 
bonding: fix miimon failure counter

During the rework of the mii monitor for:

  commit f0c76d61779b153dbfb955db3f144c62d02173c2
  Author: Jay Vosburgh <fubar@us.ibm.com>
  Date:   Wed Jul 2 18:21:58 2008 -0700

    bonding: refactor mii monitor

I left out the increment of the link failure counter.  This
patch corrects that omission.

Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
16 years agoMerge branch 'for-linus' of git://neil.brown.name/md
Linus Torvalds [Fri, 31 Oct 2008 01:36:16 +0000 (18:36 -0700)] 
Merge branch 'for-linus' of git://neil.brown.name/md

* 'for-linus' of git://neil.brown.name/md:
  md: destroy partitions and notify udev when md array is stopped.

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus
Linus Torvalds [Fri, 31 Oct 2008 01:35:09 +0000 (18:35 -0700)] 
Merge git://git./linux/kernel/git/rusty/linux-2.6-for-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
  lguest: fix irq vectors.
  lguest: fix early_ioremap.
  lguest: fix example launcher compile after moved asm-x86 dir.

16 years agoMerge branch 'x86-fixes-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 31 Oct 2008 01:33:46 +0000 (18:33 -0700)] 
Merge branch 'x86-fixes-for-linus-2' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: cpu_index build fix
  x86/voyager: fix missing cpu_index initialisation
  x86/voyager: fix compile breakage caused by dc1e35c6e95e8923cf1d3510438b63c600fee1e2
  x86: fix /dev/mem mmap breakage when PAT is disabled
  x86/voyager: fix compile breakage casued by x86: move prefill_possible_map calling early
  x86: use CONFIG_X86_SMP instead of CONFIG_SMP
  x86/voyager: fix boot breakage caused by x86: boot secondary cpus through initial_code
  x86, uv: fix compile error in uv_hub.h
  i386/PAE: fix pud_page()
  x86: remove debug code from arch_add_memory()
  x86: start annotating early ioremap pointers with __iomem
  x86: two trivial sparse annotations
  x86: fix init_memory_mapping for [dc000000 - e0000000) - v2

16 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 31 Oct 2008 01:32:03 +0000 (18:32 -0700)] 
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: remove sched-design.txt from 00-INDEX
  sched: change sched_debug's mode to 0444

16 years agoMerge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 31 Oct 2008 01:31:42 +0000 (18:31 -0700)] 
Merge branch 'tracing-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  ftrace: handle archs that do not support irqs_disabled_flags

16 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 31 Oct 2008 01:31:27 +0000 (18:31 -0700)] 
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  resources: fix x86info results ioremap.c:226 __ioremap_caller+0xf2/0x2d6() WARNINGs

16 years agolguest: fix irq vectors.
Rusty Russell [Fri, 31 Oct 2008 16:24:27 +0000 (11:24 -0500)] 
lguest: fix irq vectors.

do_IRQ: cannot handle IRQ -1 vector 0x20 cpu 0
------------[ cut here ]------------
kernel BUG at arch/x86/kernel/irq_32.c:219!

We're not ISA: we have a 1:1 mapping from vectors to irqs.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
16 years agolguest: fix early_ioremap.
Rusty Russell [Fri, 31 Oct 2008 16:24:27 +0000 (11:24 -0500)] 
lguest: fix early_ioremap.

dmi_scan_machine breaks under lguest:
lguest: unhandled trap 14 at 0xc04edeae (0xffa00000)

This is because we use current_cr3 for the read_cr3() paravirt
function, and it isn't set until the first cr3 change.  We got away
with it until this happened.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
16 years agolguest: fix example launcher compile after moved asm-x86 dir.
Rusty Russell [Fri, 31 Oct 2008 16:24:25 +0000 (11:24 -0500)] 
lguest: fix example launcher compile after moved asm-x86 dir.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
16 years agox86: cpu_index build fix
Ingo Molnar [Thu, 30 Oct 2008 23:43:03 +0000 (00:43 +0100)] 
x86: cpu_index build fix

fix:

 arch/x86/kernel/cpu/common.c: In function 'early_identify_cpu':
 arch/x86/kernel/cpu/common.c:553: error: 'struct cpuinfo_x86' has no member named 'cpu_index'

as cpu_index is only available on SMP.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86/voyager: fix missing cpu_index initialisation
James Bottomley [Thu, 30 Oct 2008 21:13:37 +0000 (16:13 -0500)] 
x86/voyager: fix missing cpu_index initialisation

Impact: fix /proc/cpuinfo output on x86/Voyager

Ever since

| commit 92cb7612aee39642d109b8d935ad265e602c0563
| Author: Mike Travis <travis@sgi.com>
| Date:   Fri Oct 19 20:35:04 2007 +0200
|
|     x86: convert cpuinfo_x86 array to a per_cpu array

We've had an extra field in cpuinfo_x86 which is cpu_index.
Unfortunately, voyager has never initialised this, although the only
noticeable impact seems to be that /proc/cpuinfo shows all zeros for
the processor ids.

Anyway, fix this by initialising the boot CPU properly and setting the
index when the secondaries update.

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86/voyager: fix compile breakage caused by dc1e35c6e95e8923cf1d3510438b63c600fee1e2
James Bottomley [Thu, 30 Oct 2008 21:00:59 +0000 (16:00 -0500)] 
x86/voyager: fix compile breakage caused by dc1e35c6e95e8923cf1d3510438b63c600fee1e2

Impact: build fix on x86/Voyager

Given commits like this:

| Author: Suresh Siddha <suresh.b.siddha@intel.com>
| Date:   Tue Jul 29 10:29:19 2008 -0700
|
|     x86, xsave: enable xsave/xrstor on cpus with xsave support

Which deliberately expose boot cpu dependence to pieces of the system,
I think it's time to explicitly have a variable for it to prevent this
continual misassumption that the boot CPU is zero.

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Thu, 30 Oct 2008 23:16:39 +0000 (16:16 -0700)] 
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 5326/1: AFEB9260: Fix for i2c_board_info structure
  [ARM] mx31ads: Add missing include
  [ARM] MXC: Fix mxc_gpio_get(), which must read PSR register instead DR.
  [ARM] MX3: Use ioremap wrapper to map SoC devices nonshared
  [ARM] gpio_free might sleep, arm architecture
  [ARM] ep93xx: fix OHCI DMA mask
  leds: da903x: (da9030 only) led brightness reversed.
  [ARM] sharpsl_pm: fix compilation w/o CONFIG_PM
  [ARM] pcm037: map AIPS1 and AIPS2 as nonshared area
  [ARM] build fixes for netX serial driver
  [ARM] 5323/1: Remove outdated empeg documentation.
  [ARM] 5299/1: Add maintainer for Mobilepro 900/c
  [ARM] corgi_lcd: fix simultaneous compilation with corgi_bl
  [ARM] pxa/spitz: fix spi cs on spitz
  [ARM] 5322/1: Fix fastpath issue in mmci.c
  [ARM] xsc3: revert writethrough memory-type encoding change

16 years agoftrace: handle archs that do not support irqs_disabled_flags
Steven Rostedt [Fri, 24 Oct 2008 13:42:59 +0000 (09:42 -0400)] 
ftrace: handle archs that do not support irqs_disabled_flags

Impact: build fix on non-lockdep architectures

Some architectures do not support a way to read the irq flags that
is set from "local_irq_save(flags)" to determine if interrupts were
disabled or enabled. Ftrace uses this information to display to the user
if the trace occurred with interrupts enabled or disabled.

Besides the fact that those archs that do not support this will fail to
compile, unless they fix it, we do not want to have the trace simply
say interrupts were not disabled or they were enabled, without knowing
the real answer.

This patch adds a 'X' in the output to let the user know that the
architecture they are running on does not support a way for the tracer
to determine if interrupts were enabled or disabled. It also lets those
same archs compile with tracing enabled.

Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: fix /dev/mem mmap breakage when PAT is disabled
Ravikiran G Thirumalai [Thu, 30 Oct 2008 20:59:21 +0000 (13:59 -0700)] 
x86: fix /dev/mem mmap breakage when PAT is disabled

Impact: allow /dev/mem mmaps on non-PAT CPUs/platforms

Fix mmap to /dev/mem when CONFIG_X86_PAT is off and CONFIG_STRICT_DEVMEM is
off

mmap to /dev/mem on kernel memory has been failing since the
introduction of PAT (CONFIG_STRICT_DEVMEM=n case).   Seems like
the check to avoid cache aliasing with PAT is kicking in even
when PAT is disabled. The bug seems to have crept in 2.6.26.

This patch makes sure that the mmap to regular
kernel memory succeeds if CONFIG_STRICT_DEVMEM=n and
PAT is disabled, and the checks to avoid cache aliasing
still happens if PAT is enabled.

Signed-off-by: Ravikiran Thirumalai <kiran@scalex86.org>
Tested-by: Tim Sirianni <tim@scalemp.com>
Cc: <stable@kernel.org>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86/voyager: fix compile breakage casued by x86: move prefill_possible_map calling...
James Bottomley [Thu, 30 Oct 2008 21:28:35 +0000 (16:28 -0500)] 
x86/voyager: fix compile breakage casued by x86: move prefill_possible_map calling early

Impact: fix build failure on x86/Voyager

Before:

| commit 329513a35d1a2b6b28d54f5c2c0dde4face8200b
| Author: Yinghai Lu <yhlu.kernel@gmail.com>
| Date:   Wed Jul 2 18:54:40 2008 -0700
|
|     x86: move prefill_possible_map calling early

prefill_possible_mask() was hidden under CONFIG_HOTPLUG_CPU rendering
it invisitble to voyager.  Since this commit it's exposed, but not
provided by the voyager subarch, so add a dummy stub to fix the link
breakage.

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: use CONFIG_X86_SMP instead of CONFIG_SMP
James Bottomley [Thu, 30 Oct 2008 21:05:39 +0000 (16:05 -0500)] 
x86: use CONFIG_X86_SMP instead of CONFIG_SMP

Impact: fix x86/Voyager boot

CONFIG_SMP is used for features which work on *all* x86 boxes.
CONFIG_X86_SMP is used for standard PC like x86 boxes (for things like
multi core and apics)

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86/voyager: fix boot breakage caused by x86: boot secondary cpus through initial_code
James Bottomley [Thu, 30 Oct 2008 21:08:38 +0000 (16:08 -0500)] 
x86/voyager: fix boot breakage caused by x86: boot secondary cpus through initial_code

Impact: boot up secondary CPUs as well on x86/Voyager systems

This commit:

| commit 3e9704739daf46a8ba6593d749c67b5f7cd633d2
| Author: Glauber Costa <gcosta@redhat.com>
| Date:   Wed May 28 13:01:54 2008 -0300
|
|     x86: boot secondary cpus through initial_code

removed the use of initialize_secondary.  However, it didn't update
voyager, so the secondary cpus no longer boot.  Fix this by adding the
initial_code switch to voyager as well.

Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Glauber Costa <gcosta@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoudp: Should use spin_lock_bh()/spin_unlock_bh() in udp_lib_unhash()
Eric Dumazet [Thu, 30 Oct 2008 21:00:53 +0000 (14:00 -0700)] 
udp: Should use spin_lock_bh()/spin_unlock_bh() in udp_lib_unhash()

Spotted by Alexander Beregalov

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes
Linus Torvalds [Thu, 30 Oct 2008 19:55:49 +0000 (12:55 -0700)] 
Merge git://git./linux/kernel/git/sam/kbuild-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes:
  Fix incompatibility with versions of Perl less than 5.6.0
  kbuild: do not include arch/<ARCH>/include/asm in find-sources twice.
  kbuild: tag with git revision when git describe is missing
  kbuild: prevent modpost from looking for a .cmd file for a static library linked into a module
  kbuild: fix KBUILD_EXTRA_SYMBOLS
  adjust init section definitions
  scripts/checksyscalls.sh: fix for non-gnu sed
  scripts/package: don't break if %{_smp_mflags} isn't set
  kbuild: setlocalversion: dont include svn change count
  kbuild: improve check-symlink
  kbuild: mkspec - fix build rpm

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Thu, 30 Oct 2008 19:54:03 +0000 (12:54 -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:
  [PATCH] Switch all my contributions stuff to a single common address
  [WATCHDOG] pci: use pci_ioremap_bar() in drivers/watchdog

16 years agoMerge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus
Linus Torvalds [Thu, 30 Oct 2008 19:53:33 +0000 (12:53 -0700)] 
Merge branch 'upstream' of git://ftp.linux-mips.org/upstream-linus

* 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus:
  CHAR: Delete old and now unused M48T35 RTC driver for SGI IP27.
  CHAR: Delete old and now unused DS1286 driver.
  MIPS: Sort out CPU type to name translation.
  MIPS: Use the new byteorder headers
  MIPS: Probe for watch registers on cores of all vendors, not just MTI.
  MIPS: Switch FPU emulator trap to BREAK instruction.
  MIPS: SMP: Do not initialize __cpu_number_map/__cpu_logical_map for CPU 0.
  MIPS: Consider value of c0_ebase when computing value of exception base.
  MIPS: Clean up MIPSxx-optimized bitop functions
  MIPS: New feature test macro cpu_has_mips_r
  MIPS: RBTX4927: Add GPIO-LED support
  MIPS: TXx9: Fix RBTX4939 ethernet address initialization

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/brodo/pcmcia-2.6
Linus Torvalds [Thu, 30 Oct 2008 19:52:53 +0000 (12:52 -0700)] 
Merge git://git./linux/kernel/git/brodo/pcmcia-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/brodo/pcmcia-2.6:
  fdomain_cs: Sort out modules with duplicate description
  pcmcia: Whine harder about use of EXCLUSIVE
  pcmcia: IRQ_TYPE_EXCLUSIVE is long obsoleted

16 years agoMerge branch 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6
Linus Torvalds [Thu, 30 Oct 2008 19:51:42 +0000 (12:51 -0700)] 
Merge branch 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* 'bugfixes' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6:
  SUNRPC: Fix potential race in put_rpccred()
  SUNRPC: Fix rpcauth_prune_expired
  NFS: Convert nfs_attr_generation_counter into an atomic_long
  SUNRPC: Respond promptly to server TCP resets

16 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 30 Oct 2008 19:50:59 +0000 (12:50 -0700)] 
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, gart: fix gart detection for Fam11h CPUs
  x86: 64 bit print out absent pages num too
  x86, kdump: fix invalid access on i386 sparsemem
  x86: fix APIC_DEBUG with inquire_remote_apic
  x86: AMD microcode patch loader author update
  x86: microcode patch loader author update
  mailmap: add Peter Oruba
  x86, bts: improve help text for BTS config
  doc/x86: fix doc subdirs

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm
Linus Torvalds [Thu, 30 Oct 2008 19:16:38 +0000 (12:16 -0700)] 
Merge git://git./linux/kernel/git/agk/linux-2.6-dm

* git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm:
  dm snapshot: wait for chunks in destructor
  dm snapshot: fix register_snapshot deadlock
  dm raid1: fix do_failures

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Thu, 30 Oct 2008 19:09:44 +0000 (12:09 -0700)] 
Merge branch 'for-linus' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  x86/PCI: build failure at x86/kernel/pci-dma.c with !CONFIG_PCI

16 years agotty: Fix USB kref leak
Alan Cox [Thu, 30 Oct 2008 15:54:12 +0000 (15:54 +0000)] 
tty: Fix USB kref leak

When we close we must clear the extra reference we got when we read
port->tty. Setting the port tty NULL will clear the kref held by the driver
but not the one we obtained ourselves while doing the lookup.

Signed-off-by: Alan Cox <alan@redhat.com>
Tested-by: Helge Hafting <helge.hafting@aitel.hist.no>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6
Linus Torvalds [Thu, 30 Oct 2008 19:07:25 +0000 (12:07 -0700)] 
Merge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6

* 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6:
  i2c-s3c2410: Correct use of ! and &
  i2c: The i2c mailing list is moving
  scx200_i2c: Add missing class parameter