firefly-linux-kernel-4.4.55.git
9 years agomac80211: remove 80+80 MHz rate reporting
Johannes Berg [Thu, 15 Jan 2015 15:02:46 +0000 (16:02 +0100)]
mac80211: remove 80+80 MHz rate reporting

These rates are treated the same as 160 MHz in the spec,
so it makes no sense to distinguish them. As no driver
uses them yet, this is also not a problem, just remove
them.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: docs: remove station_info_flags
Johannes Berg [Wed, 14 Jan 2015 12:57:36 +0000 (13:57 +0100)]
cfg80211: docs: remove station_info_flags

I removed the enum earlier, but forgot to remove it from the
documentation - do that now.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: send netdetect configuration info in NL80211_CMD_GET_WOWLAN
Luciano Coelho [Fri, 9 Jan 2015 12:06:37 +0000 (14:06 +0200)]
nl80211: send netdetect configuration info in NL80211_CMD_GET_WOWLAN

Send the netdetect configuration information in the response to
NL8021_CMD_GET_WOWLAN commands.  This includes the scan interval,
SSIDs to match and frequencies to scan.

Additionally, add the NL80211_WOWLAN_TRIG_NET_DETECT with
NL80211_ATTR_WOWLAN_TRIGGERS_SUPPORTED.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: avoid reg-hints in self-managed only systems
Arik Nemtsov [Wed, 7 Jan 2015 14:47:20 +0000 (16:47 +0200)]
cfg80211: avoid reg-hints in self-managed only systems

When a system contains only self-managed regulatory devices all hints
from the regulatory core are ignored. Stop hint processing early in this
case. These systems usually don't have CRDA deployed, which results in
endless (irrelevent) logs of the form:
cfg80211: Calling CRDA to update world regulatory domain

Make sure there's at least one self-managed device before discarding a
hint, in order to prevent initial hints from disappearing on CRDA
managed systems.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: introduce sync regdom set API for self-managed
Arik Nemtsov [Wed, 7 Jan 2015 14:47:19 +0000 (16:47 +0200)]
cfg80211: introduce sync regdom set API for self-managed

A self-managed device will sometimes need to set its regdomain synchronously.
Notably it should be set before usermode has a chance to query it. Expose
a new API to accomplish this which requires the RTNL.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: don't defer scans in case of radar detection
Eliad Peller [Wed, 7 Jan 2015 15:50:11 +0000 (17:50 +0200)]
mac80211: don't defer scans in case of radar detection

Radar detection can last indefinite time. There is no
point in deferring a scan request in this case - simply
return -EBUSY.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: consider only relevant vifs for radar_required calculation
Eliad Peller [Wed, 7 Jan 2015 15:50:10 +0000 (17:50 +0200)]
mac80211: consider only relevant vifs for radar_required calculation

ctx->conf.radar_enabled should reflect whether radar
detection is enabled for the channel context.

When calculating it, make it consider only the vifs
that have this context assigned (instead of all the
vifs).

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: remove local->radar_detect_enabled
Eliad Peller [Wed, 7 Jan 2015 15:50:09 +0000 (17:50 +0200)]
mac80211: remove local->radar_detect_enabled

local->radar_detect_enabled should tell whether
radar_detect is enabled on any interface belonging
to local.

However, it's not getting updated correctly
in many cases (actually, when testing with hwsim
it's never been set, even when the dfs master
is beaconing).

Instead of handling all the corner cases
(e.g. channel switch), simply check whether
radar detection is enabled only when needed,
instead of caching the result.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: add TDLS supported channels correctly
Arik Nemtsov [Wed, 7 Jan 2015 14:45:07 +0000 (16:45 +0200)]
mac80211: add TDLS supported channels correctly

The function adding the supported channels IE during a TDLS connection had
several issues:
1. If the entire subband is usable, the function exitted the loop without
   adding it
2. The function only checked chandef_usable, ignoring flags like RADAR
   which would prevent TDLS off-channel communcation.
3. HT20 was explicitly required in the chandef, while not a requirement
   for TDLS off-channel.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: let flush() drop packets when possible
Emmanuel Grumbach [Wed, 7 Jan 2015 13:42:39 +0000 (15:42 +0200)]
mac80211: let flush() drop packets when possible

When roaming / suspending, it makes no sense to wait until
the transmit queues of the device are empty. In extreme
condition they can be starved (VO saturating the air), but
even in regular cases, it is pointless to delay the roaming
because the low level driver is trying to send packets to
an AP which is far away. We'd rather drop these packets and
let TCP retransmit if needed. This will allow to speed up
the roaming.

For suspend, the explanation is even more trivial.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agorfkill: document rfkill module parameters
Andrew Clausen [Fri, 9 Jan 2015 20:24:55 +0000 (20:24 +0000)]
rfkill: document rfkill module parameters

Document the rfkill module parameters default_state and
master_switch_mode.

Signed-off-by: Andrew Clausen <andrew.p.clausen@gmail.com>
[rewrite commit message]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: fix handling TIM IE when stations disconnect
Johannes Berg [Fri, 9 Jan 2015 10:40:39 +0000 (11:40 +0100)]
mac80211: fix handling TIM IE when stations disconnect

When a station disconnects with frames still pending, we clear
the TIM bit, but too late - it's only cleared when the station
is already removed from the driver, and thus the driver can get
confused (and hwsim will loudly complain.)

Fix this by clearing the TIM bit earlier, when the station has
been unlinked but not removed from the driver yet. To do this,
refactor the TIM recalculation to in that case ignore traffic
and simply assume no pending traffic - this is correct for the
disconnected station even though the frames haven't been freed
yet at that point.

This patch isn't needed for current drivers though as they don't
check the station argument to the set_tim() operation and thus
don't really run into the possible confusion.

Reported-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211_hwsim: fix PS debugfs file locking
Johannes Berg [Fri, 9 Jan 2015 10:29:50 +0000 (11:29 +0100)]
mac80211_hwsim: fix PS debugfs file locking

The functions called within the iterators must be called with
tasklets disabled, so use atomic iteration like the rest of
the code and disable tasklets around the whole operation.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: provide per-TID RX/TX MSDU counters
Johannes Berg [Fri, 21 Nov 2014 13:26:31 +0000 (14:26 +0100)]
mac80211: provide per-TID RX/TX MSDU counters

Implement the new counters cfg80211 can now advertise to userspace.
The TX code is in the sequence number handler, which is a bit odd,
but that place already knows the TID and frame type, so it was
easiest and least impact there.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: support per-TID station statistics
Johannes Berg [Fri, 19 Dec 2014 11:34:00 +0000 (12:34 +0100)]
nl80211: support per-TID station statistics

The base for the current statistics is pretty mixed up, support
exporting RX/TX statistics for MSDUs per TID. This (currently)
covers received MSDUs, transmitted MSDUs and retries/failures
thereof.

Doing it per TID for MSDUs makes more sense than say only per AC
because it's symmetric - we could export per-AC statistics for all
frames (which AC we used for transmission can be determined also
for management frames) but per TID is better and usually data
frames are really the ones we care about. Also, on RX we can't
determine the AC - but we do know the TID for any QoS MPDU we
received.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: clarify packet statistics descriptions
Johannes Berg [Fri, 21 Nov 2014 11:40:05 +0000 (12:40 +0100)]
nl80211: clarify packet statistics descriptions

The current statistics we keep aren't very clear, some are on
MPDUs and some on MSDUs/MMPDUs. Clarify the descriptions based
on the counters mac80211 keeps.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: add nl80211 beacon-only statistics
Johannes Berg [Mon, 17 Nov 2014 13:12:22 +0000 (14:12 +0100)]
cfg80211: add nl80211 beacon-only statistics

Add these two values:
 * BEACON_RX: number of beacons received from this peer
 * BEACON_SIGNAL_AVG: signal strength average for beacons only

These can then be used for Android Lollipop's statistics request.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: remove enum station_info_flags
Johannes Berg [Mon, 17 Nov 2014 13:08:11 +0000 (14:08 +0100)]
cfg80211: remove enum station_info_flags

This is really just duplicating the list of information that's
already available in the nl80211 attribute, so remove the list.
Two small changes are needed:
 * remove STATION_INFO_ASSOC_REQ_IES complete, but the length
   (assoc_req_ies_len) can be used instead
 * add NL80211_STA_INFO_RX_DROP_MISC which exists internally
   but not in nl80211 yet

This gets rid of the duplicate maintenance of the two lists.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: allow drivers to provide most station statistics
Johannes Berg [Mon, 17 Nov 2014 10:35:23 +0000 (11:35 +0100)]
mac80211: allow drivers to provide most station statistics

In many cases, drivers can filter things like beacons that will
skew statistics reported by mac80211. To get correct statistics
in these cases, call drivers to obtain statistics and let them
override all values, filling values from mac80211 if the driver
didn't provide them. Not all of them make sense for the driver
to fill, so some are still always done by mac80211.

Note that this doesn't currently allow a driver to say "I know
this value is wrong, don't report it at all", or to sum it up
with a mac80211 value (as could be useful for "dropped misc"),
that can be added if it turns out to be needed.

This also gets rid of the get_rssi() method as is can now be
implemented using sta_statistics().

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: send statistics with delete station event
Johannes Berg [Fri, 14 Nov 2014 19:32:57 +0000 (20:32 +0100)]
mac80211: send statistics with delete station event

Use the new cfg80211_del_sta_sinfo() function to send the
statistics about the deleted station with the delete event.
This lets userspace see how much traffic etc. the deleted
station used.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: allow including station info in delete event
Johannes Berg [Fri, 14 Nov 2014 16:14:00 +0000 (17:14 +0100)]
cfg80211: allow including station info in delete event

When a station is removed, its statistics may be interesting to
userspace, for example for further aggregation of statistics of
all stations that ever connected to an AP.

Introduce a new cfg80211_del_sta_sinfo() function (and make the
cfg80211_del_sta() a static inline calling it) to allow passing
a struct station_info along with this, and send the data in the
nl80211 event message.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: add scan time to survey data
Johannes Berg [Fri, 14 Nov 2014 15:44:11 +0000 (16:44 +0100)]
cfg80211: add scan time to survey data

Add the time spent scanning to the survey data so it can be
reported by drivers that collect such information.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: allow survey data to return global data
Johannes Berg [Fri, 14 Nov 2014 15:43:50 +0000 (16:43 +0100)]
cfg80211: allow survey data to return global data

Not all devices are able to report survey data (particularly
time spent for various operations) per channel. As all these
statistics already exist in survey data, allow such devices
to report them (if userspace requested it)

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: remove "channel" from survey names
Johannes Berg [Fri, 14 Nov 2014 15:35:34 +0000 (16:35 +0100)]
cfg80211: remove "channel" from survey names

All of the survey data is (currently) per channel anyway,
so having the word "channel" in the name does nothing. In
the next patch I'll introduce global data to the survey,
where the word "channel" is actually confusing.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: Re-fix accounting of the tailroom-needed counter
Ido Yariv [Tue, 6 Jan 2015 13:39:02 +0000 (08:39 -0500)]
mac80211: Re-fix accounting of the tailroom-needed counter

When hw acceleration is enabled, the GENERATE_IV or PUT_IV_SPACE flags
only require headroom space. Therefore, the tailroom-needed counter can
safely be decremented for most drivers.

The older incarnation of this patch (ca34e3b5) assumed that the above
holds true for all drivers. As reported by Christopher Chavez and
researched by Christian Lamparter and Larry Finger, this isn't a valid
assumption for p54 and cw1200.

Drivers that still require tailroom for ICV/MIC even when HW encryption
is enabled can use IEEE80211_KEY_FLAG_RESERVE_TAILROOM to indicate it.

Signed-off-by: Ido Yariv <idox.yariv@intel.com>
Cc: Christopher Chavez <chrischavez@gmx.us>
Cc: Christian Lamparter <chunkeey@googlemail.com>
Cc: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoMerge branch 'mac80211' into mac80211-next
Johannes Berg [Wed, 7 Jan 2015 13:39:02 +0000 (14:39 +0100)]
Merge branch 'mac80211' into mac80211-next

Merge mac80211.git to get some changes that would otherwise
cause conflicts with new changes coming here.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: skip disabled channels in VHT check
Arik Nemtsov [Thu, 1 Jan 2015 11:43:17 +0000 (13:43 +0200)]
mac80211: skip disabled channels in VHT check

The patch "40a11ca mac80211: check if channels allow 80 MHz for VHT
probe requests" considered disabled channels as VHT enabled, and
mistakenly sent out probe-requests with the VHT IE.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: define multicast group names in header
Johannes Berg [Tue, 23 Dec 2014 16:17:38 +0000 (17:17 +0100)]
nl80211: define multicast group names in header

Put the group names into the userspace API header file so that
userspace clients can use symbolic names from there instead of
hardcoding the actual names. This doesn't really change much,
but seems somewhat cleaner.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: add extensible feature flag attribute
Gautam Kumar Shukla [Tue, 23 Dec 2014 15:55:19 +0000 (16:55 +0100)]
cfg80211: add extensible feature flag attribute

With the wiphy::features flag being used up this patch adds a
new field wiphy::ext_features. Considering extensibility this
new field is declared as a byte array. This extensible flag is
exposed to user-space by NL80211_ATTR_EXT_FEATURES.

Cc: Avinash Patil <patila@marvell.com>
Signed-off-by: Gautam (Gautam Kumar) Shukla <gautams@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: document NL80211_BSS_STATUS_AUTHENTICATED isn't used
Johannes Berg [Mon, 5 Jan 2015 10:16:42 +0000 (11:16 +0100)]
nl80211: document NL80211_BSS_STATUS_AUTHENTICATED isn't used

The flag is no longer used (and hasn't been for a long time)
since trying to track authentication (and make decisions based
on state) was just causing issues all over - see commit
95de817b9034d50860319f6033ec85d25024694c.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoRevert "mac80211: Fix accounting of the tailroom-needed counter"
Johannes Berg [Mon, 5 Jan 2015 09:28:49 +0000 (10:28 +0100)]
Revert "mac80211: Fix accounting of the tailroom-needed counter"

This reverts commit ca34e3b5c808385b175650605faa29e71e91991b.

It turns out that the p54 and cw2100 drivers assume that there's
tailroom even when they don't say they really need it. However,
there's currently no way for them to explicitly say they do need
it, so for now revert this.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=90331.

Cc: stable@vger.kernel.org
Fixes: ca34e3b5c808 ("mac80211: Fix accounting of the tailroom-needed counter")
Reported-by: Christopher Chavez <chrischavez@gmx.us>
Bisected-by: Larry Finger <Larry.Finger@lwfinger.net>
Debugged-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: fix dot11MulticastTransmittedFrameCount tested address
Eliad Peller [Sun, 21 Dec 2014 13:25:28 +0000 (15:25 +0200)]
mac80211: fix dot11MulticastTransmittedFrameCount tested address

dot11MulticastTransmittedFrameCount should be updated according
to the DA, which might be different from A1. Checking A1 results
in the counter being 0 in case of station, as to-DS data frames
use A1 for the BSSID.

This behaviour is defined in state machines, specifically in the
sta_tx_dcf_3.1d(10) description of 802.11-2012.

Signed-off-by: Eliad Peller <eliad@wizery.com>
[rewrite commit message]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211_hwsim: fix check for custom world regdom array size
Johannes Berg [Mon, 22 Dec 2014 11:51:25 +0000 (12:51 +0100)]
mac80211_hwsim: fix check for custom world regdom array size

David Binderman reports that the conditions in the first loop
are the wrong way around - checking the array contents before
the size.

Instead of leaving the empty loop there and reordering the two
checks unify it into a single loop that skips over non-matches
and exits after the first match.

Reported-by: David Binderman <dcb314@hotmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: minstrel: reduce size of struct minstrel_rate_stats
Felix Fietkau [Wed, 17 Dec 2014 12:38:34 +0000 (13:38 +0100)]
mac80211: minstrel: reduce size of struct minstrel_rate_stats

On minstrel_ht, the size of the per-sta struct is almost 18k, making it
an order-3 allocation.
A few fields inside the per-rate statistics are bigger than they need to
be. This patch reduces the size enough to cut down the per-sta struct to
about 13k (order-2 allocation).

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: increase the max number of rules in regdomain
Arik Nemtsov [Wed, 17 Dec 2014 16:00:44 +0000 (18:00 +0200)]
nl80211: increase the max number of rules in regdomain

Some network cards (Intel) produce per-channel regdomains and rely on
cfg80211 to merge rules as needed. This hits the max rules limit and
fails.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: Stop scheduled scan if netlink client disappears
Jukka Rissanen [Mon, 15 Dec 2014 11:25:39 +0000 (13:25 +0200)]
nl80211: Stop scheduled scan if netlink client disappears

An attribute NL80211_ATTR_SOCKET_OWNER can be set by the scan initiator.
If present, the attribute will cause the scan to be stopped if the client
dies.

Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: Convert sched_scan_req pointer to RCU pointer
Jukka Rissanen [Mon, 15 Dec 2014 11:25:38 +0000 (13:25 +0200)]
nl80211: Convert sched_scan_req pointer to RCU pointer

Because of possible races when accessing sched_scan_req pointer in
rdev, the sched_scan_req is converted to RCU pointer.

Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: avoid reconfig if no interfaces are up
Eliad Peller [Sun, 14 Dec 2014 09:05:53 +0000 (11:05 +0200)]
mac80211: avoid reconfig if no interfaces are up

If there are no interfaces up, there is no reason
to continue the reconfig flow.

The current code might end up calling driver
callbacks (e.g. resume(), reconfig_complete())
while the driver is already stopped.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: remove unused variable in ieee80211_parse_ch_switch_ie()
Luciano Coelho [Sun, 14 Dec 2014 09:05:52 +0000 (11:05 +0200)]
mac80211: remove unused variable in ieee80211_parse_ch_switch_ie()

The ht_oper variable is assigned a value, but never used in
ieee80211_parse_ch_switch_ie().  Remove it.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: update sta bw on ht chanwidth action frame
Eliad Peller [Sun, 14 Dec 2014 09:05:51 +0000 (11:05 +0200)]
mac80211: update sta bw on ht chanwidth action frame

Commit e1a0c6b ("mac80211: stop toggling IEEE80211_HT_CAP_SUP_WIDTH_20_40")
mistakenly removed the actual update of sta->sta.bandwidth.

Refactor ieee80211_sta_cur_vht_bw() into multiple functions
(calculate caps-bw and chandef-bw separately, and min them
with cur_max_bandwidth).

On ht chanwidth action frame set only cur_max_bandwidth
(according to the sta capabilities) and recalc the sta bw.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: handle power constraint and country IEs in RRM
Moshe Benji [Sun, 14 Dec 2014 09:05:50 +0000 (11:05 +0200)]
mac80211: handle power constraint and country IEs in RRM

In beacons, handle the Country IE even if no Power Constraint IE
is present, and, capability wise, also in case that the Radio
Measurements capability is enabled.

In cases where the Country IE should be handled and that the
Power Constraint IE is not present, the Country IE alone will
set the power limit (and not both Country and Power Constraint
IEs).

Signed-off-by: Moshe Benji <moshe.benji@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: Fix ignored HT override configurations
Chaya Rachel Ivgi [Sun, 14 Dec 2014 08:38:59 +0000 (10:38 +0200)]
mac80211: Fix ignored HT override configurations

HT override configurations was ignored when choosing the channel
(until now, the override configuration affected only the
capabilities shown in the IEs).

The override configurations received only on association time,
so in this case we should determine the channel again.

Signed-off-by: Chaya Rachel Ivgi <chaya.rachel.ivgi@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: free management frame keys when removing station
Johannes Berg [Wed, 17 Dec 2014 12:55:49 +0000 (13:55 +0100)]
mac80211: free management frame keys when removing station

When writing the code to allow per-station GTKs, I neglected to
take into account the management frame keys (index 4 and 5) when
freeing the station and only added code to free the first four
data frame keys.

Fix this by iterating the array of keys over the right length.

Cc: stable@vger.kernel.org
Fixes: e31b82136d1a ("cfg80211/mac80211: allow per-station GTKs")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: avoid intersection when applying self-managed reg
Arik Nemtsov [Mon, 15 Dec 2014 17:26:02 +0000 (19:26 +0200)]
cfg80211: avoid intersection when applying self-managed reg

The custom-reg handling function can currently only add flags to a given
channel. This results in stale flags being left applied. In some cases
a channel was disabled and even the orig_flags were changed to reflect
this.

Previously the API was designed for a single invocation before wiphy
registration, so this didn't matter. The previous approach doesn't scale
well to self-managed regulatory devices, particularly when a more
permissive regdom is applied after a restrictive one.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: return private regdom for self-managed devices
Arik Nemtsov [Mon, 15 Dec 2014 17:26:01 +0000 (19:26 +0200)]
cfg80211: return private regdom for self-managed devices

If a device has self-managed regulatory, insist on returning the wiphy
specific regdomain if a wiphy-idx is specified. The global regdomain is
meaningless for such devices.

Also add an attribute for self-managed devices, so usermode can
distinguish them as such.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: allow wiphy specific regdomain management
Jonathan Doron [Mon, 15 Dec 2014 17:26:00 +0000 (19:26 +0200)]
cfg80211: allow wiphy specific regdomain management

Add a new regulatory flag that allows a driver to manage regdomain
changes/updates for its own wiphy.
A self-managed wiphys only employs regulatory information obtained from
the FW and driver and does not use other cfg80211 sources like
beacon-hints, country-code IEs and hints from other devices on the same
system. Conversely, a self-managed wiphy does not share its regulatory
hints with other devices in the system. If a system contains several
devices, one or more of which are self-managed, there might be
contradictory regulatory settings between them. Usage of flag is
generally discouraged. Only use it if the FW/driver is incompatible
with non-locally originated hints.

A new API lets the driver send a complete regdomain, to be applied on
its wiphy only.

After a wiphy-specific regdomain change takes place, usermode will get
a new type of change notification. The regulatory core also takes care
enforce regulatory restrictions, in case some interfaces are on
forbidden channels.

Signed-off-by: Jonathan Doron <jonathanx.doron@intel.com>
Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: allow usermode to query wiphy specific regdom
Arik Nemtsov [Mon, 15 Dec 2014 17:25:59 +0000 (19:25 +0200)]
cfg80211: allow usermode to query wiphy specific regdom

If a wiphy-idx is specified, the kernel will return the wiphy specific
regdomain, if such exists. Otherwise return the global regdom.

When no wiphy-idx is specified, return the global regdomain as well as
all wiphy-specific regulatory domains in the system, via a new nested
list of attributes.

Add a new attribute for each wiphy-specific regdomain, for usermode to
identify it as such.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: keep sending peer candidate events while in listen state
Nishikawa, Kenzoh [Tue, 16 Dec 2014 01:41:06 +0000 (01:41 +0000)]
mac80211: keep sending peer candidate events while in listen state

Instead of sending peer candidate events just once, send them as long
as the peer remains in the LISTEN state in the peering state machine,
when userspace is implementing the peering manager.
Userspace may silence the events from a peer by progressing the state
machine or by setting the link state to BLOCKED.

Fixes the problem that a mesh peering process won't be fired again after
the previous first peering trial fails due to like air propagation error
if the peering is managed by user space such as wpa_supplicant.

This patch works with another patch for wpa_supplicant described here
which fires a peering process again triggered by the notice from kernel.
http://lists.shmoo.com/pipermail/hostap/2014-November/031235.html

Signed-off-by: Kenzoh Nishikawa <Kenzoh.Nishikawa@jp.sony.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: notify channel switch at the end of ieee80211_chswitch_post_beacon()
Luciano Coelho [Tue, 16 Dec 2014 14:01:39 +0000 (16:01 +0200)]
mac80211: notify channel switch at the end of ieee80211_chswitch_post_beacon()

The call to cfg80211_ch_switch_notify() should be at the end of the
ieee80211_chswitch_post_beacon() function, because it should only be
sent if everything succeeded.

Fixes: d04b5ac9e70b ("cfg80211/mac80211: allow any interface to send channel switch notifications")
Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: notify NSS changed when IBSS and HT
Janusz Dziedzic [Wed, 17 Dec 2014 09:00:52 +0000 (10:00 +0100)]
mac80211: notify NSS changed when IBSS and HT

When using IBSS in HT mode, we always get NSS=1
in rc_update callback. Force NSS recalculation when
rates updated and notify driver that NSS changed.

Signed-off-by: Janusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: move U-APSD enablement to vif flags
Johannes Berg [Tue, 11 Nov 2014 11:48:42 +0000 (12:48 +0100)]
mac80211: move U-APSD enablement to vif flags

In order to let drivers have more dynamic U-APSD support,
move the enablement flag to the virtual interface driver
flags. This lets drivers not only set it up differently
for different interfaces, but also enable/disable on the
fly if needed.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: ask driver to look at power level when starting AP
Johannes Berg [Thu, 13 Nov 2014 10:23:53 +0000 (11:23 +0100)]
mac80211: ask driver to look at power level when starting AP

The power level might have been set, but as the interface was idle
it might not have taken effect yet. Ask the driver to check the
power level when starting up an AP so that in this case the correct
power level is used in case the device/driver can only set it when
the interface is actually active.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: Fix accounting of multicast frames
Sujith Manoharan [Wed, 10 Dec 2014 15:56:11 +0000 (21:26 +0530)]
mac80211: Fix accounting of multicast frames

Since multicast frames are marked as no-ack, using
IEEE80211_TX_STAT_ACK to check if they have been
successfully transmitted by the driver is incorrect
since a driver can choose to ignore transmission status
for no-ack frames. This results in incorrect accounting
for such frames.

To fix this issue, this patch introduces a new flag
that can be used by drivers to indicate error-free
transmission of no-ack frames.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
[add a note about not setting the flag for non-no-ack frames]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: Move IEEE80211_TX_CTL_PS_RESPONSE
Sujith Manoharan [Wed, 10 Dec 2014 15:56:10 +0000 (21:26 +0530)]
mac80211: Move IEEE80211_TX_CTL_PS_RESPONSE

Move IEEE80211_TX_CTL_PS_RESPONSE to info->control.flags since
this is used only in the TX path (by ath9k). This frees up
a bit which can be used for other purposes.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agowireless: Support of IFLA_INFO_KIND rtnl attribute
Vadim Kochan [Tue, 9 Dec 2014 22:21:06 +0000 (00:21 +0200)]
wireless: Support of IFLA_INFO_KIND rtnl attribute

It allows to identify the wlan kind of device for the user application,
e.g.:

    # ip -d link

    1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default
        link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 promiscuity 0
    2: enp0s25: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast state DOWN mode DEFAULT group default qlen 1000
        link/ether XX:XX:XX:XX:XX:XX brd ff:ff:ff:ff:ff:ff promiscuity 0
    3: wlp3s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
        link/ether XX:XX:XX:XX:XX:XX brd ff:ff:ff:ff:ff:ff promiscuity 0
        wlan

Signed-off-by: Vadim Kochan <vadim4j@gmail.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
[make wireless_link_ops const]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: correctly check ad-hoc channels
Arik Nemtsov [Wed, 3 Dec 2014 16:08:17 +0000 (18:08 +0200)]
cfg80211: correctly check ad-hoc channels

Ad-hoc requires beaconing for regulatory purposes. Validate that the
channel is valid for beaconing, and not only enabled.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: don't WARN about two consecutive Country IE hint
Emmanuel Grumbach [Tue, 2 Dec 2014 07:53:25 +0000 (09:53 +0200)]
cfg80211: don't WARN about two consecutive Country IE hint

This can happen and there is no point in added more
detection code lower in the stack. Catching these in one
single point (cfg80211) is enough. Stop WARNING about this
case.

This fixes:
https://bugzilla.kernel.org/show_bug.cgi?id=89001

Cc: stable@vger.kernel.org
Fixes: 2f1c6c572d7b ("cfg80211: process non country IE conflicting first")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: update the channel context after channel switch
Emmanuel Grumbach [Sun, 30 Nov 2014 15:17:26 +0000 (17:17 +0200)]
mac80211: update the channel context after channel switch

When the channel switch has been made, a vif is now using
the channel context which was reserved. When that happens,
we need to update the channel context since its parameters
may change.

I hit a case in which I switched to a 40Mhz channel but the
reserved channel context was still on 20Mhz. The rate control
would try to send 40Mhz packets on a 20Mhz channel context and
that made iwlwifi's firmware unhappy.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: check matches array length before acessing it
Luciano Coelho [Mon, 1 Dec 2014 09:32:09 +0000 (11:32 +0200)]
nl80211: check matches array length before acessing it

If the userspace passes a malformed sched scan request (or a net
detect wowlan configuration) by adding a NL80211_ATTR_SCHED_SCAN_MATCH
attribute without any nested matchsets, a NULL pointer dereference
will occur.  Fix this by checking that we do have matchsets in our
array before trying to access it.

BUG: unable to handle kernel NULL pointer dereference at 0000000000000024
IP: [<ffffffffa002fd69>] nl80211_parse_sched_scan.part.67+0x6e9/0x900 [cfg80211]
PGD 865c067 PUD 865b067 PMD 0
Oops: 0002 [#1] SMP
Modules linked in: iwlmvm(O) iwlwifi(O) mac80211(O) cfg80211(O) compat(O) [last unloaded: compat]
CPU: 2 PID: 2442 Comm: iw Tainted: G           O   3.17.2 #31
Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
task: ffff880013800790 ti: ffff880008d80000 task.ti: ffff880008d80000
RIP: 0010:[<ffffffffa002fd69>]  [<ffffffffa002fd69>] nl80211_parse_sched_scan.part.67+0x6e9/0x900 [cfg80211]
RSP: 0018:ffff880008d838d0  EFLAGS: 00010293
RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 000000000000143c RSI: 0000000000000000 RDI: ffff880008ee8dd0
RBP: ffff880008d83948 R08: 0000000000000002 R09: 0000000000000019
R10: ffff88001d1b3c40 R11: 0000000000000002 R12: ffff880019e85e00
R13: 00000000fffffed4 R14: ffff880009757800 R15: 0000000000001388
FS:  00007fa3b6d13700(0000) GS:ffff88003e200000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000024 CR3: 0000000008670000 CR4: 00000000000006e0
Stack:
 ffff880009757800 ffff880000000001 0000000000000000 ffff880008ee84e0
 0000000000000000 ffff880009757800 00000000fffffed4 ffff880008d83948
 ffffffff814689c9 ffff880009757800 ffff880008ee8000 0000000000000000
Call Trace:
 [<ffffffff814689c9>] ? nla_parse+0xb9/0x120
 [<ffffffffa00306de>] nl80211_set_wowlan+0x75e/0x960 [cfg80211]
 [<ffffffff810bf3d5>] ? mark_held_locks+0x75/0xa0
 [<ffffffff8161a77b>] genl_family_rcv_msg+0x18b/0x360
 [<ffffffff810bf66d>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8161a9d4>] genl_rcv_msg+0x84/0xc0
 [<ffffffff8161a950>] ? genl_family_rcv_msg+0x360/0x360
 [<ffffffff81618e79>] netlink_rcv_skb+0xa9/0xd0
 [<ffffffff81619458>] genl_rcv+0x28/0x40
 [<ffffffff816184a5>] netlink_unicast+0x105/0x180
 [<ffffffff8161886f>] netlink_sendmsg+0x34f/0x7a0
 [<ffffffff8105a097>] ? kvm_clock_read+0x27/0x40
 [<ffffffff815c644d>] sock_sendmsg+0x8d/0xc0
 [<ffffffff811a75c9>] ? might_fault+0xb9/0xc0
 [<ffffffff811a756e>] ? might_fault+0x5e/0xc0
 [<ffffffff815d5d26>] ? verify_iovec+0x56/0xe0
 [<ffffffff815c73e0>] ___sys_sendmsg+0x3d0/0x3e0
 [<ffffffff810a7be8>] ? sched_clock_cpu+0x98/0xd0
 [<ffffffff810611b4>] ? __do_page_fault+0x254/0x580
 [<ffffffff810bb39f>] ? up_read+0x1f/0x40
 [<ffffffff810611b4>] ? __do_page_fault+0x254/0x580
 [<ffffffff812146ed>] ? __fget_light+0x13d/0x160
 [<ffffffff815c7b02>] __sys_sendmsg+0x42/0x80
 [<ffffffff815c7b52>] SyS_sendmsg+0x12/0x20
 [<ffffffff81751f69>] system_call_fastpath+0x16/0x1b

Fixes: ea73cbce4e1f ("nl80211: fix scheduled scan RSSI matchset attribute confusion")
Cc: stable@vger.kernel.org [3.15+]
Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: use __force __rcu to suppress sparse warning
Johannes Berg [Fri, 12 Dec 2014 11:26:25 +0000 (12:26 +0100)]
cfg80211: use __force __rcu to suppress sparse warning

The code assigns a constant value (a pointer to a static variable)
to an RCU pointer, which results in a sparse warning:
  reg.c:112:10: warning: cast adds address space to expression (<asn:4>)

Suppress this warning by using __force.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: avoid mem leak on driver hint set
Arik Nemtsov [Thu, 4 Dec 2014 10:22:16 +0000 (12:22 +0200)]
cfg80211: avoid mem leak on driver hint set

In the already-set and intersect case of a driver-hint, the previous
wiphy regdomain was not freed before being reset with a copy of the
cfg80211 regdomain.

Cc: stable@vger.kernel.org
Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Acked-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: Fix 160 MHz channels with 80+80 and 160 MHz drivers
Jouni Malinen [Thu, 11 Dec 2014 21:48:55 +0000 (23:48 +0200)]
cfg80211: Fix 160 MHz channels with 80+80 and 160 MHz drivers

The VHT supported channel width field is a two bit integer, not a
bitfield. cfg80211_chandef_usable() was interpreting it incorrectly and
ended up rejecting 160 MHz channel width if the driver indicated support
for both 160 and 80+80 MHz channels.

Cc: stable@vger.kernel.org (3.16+)
Fixes: 3d9d1d6656a73 ("nl80211/cfg80211: support VHT channel configuration")
       (however, no real drivers had 160 MHz support it until 3.16)
Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: fix multicast LED blinking and counter
Andreas Müller [Fri, 12 Dec 2014 11:11:11 +0000 (12:11 +0100)]
mac80211: fix multicast LED blinking and counter

As multicast-frames can't be fragmented, "dot11MulticastReceivedFrameCount"
stopped being incremented after the use-after-free fix. Furthermore, the
RX-LED will be triggered by every multicast frame (which wouldn't happen
before) which wouldn't allow the LED to rest at all.

Fixes https://bugzilla.kernel.org/show_bug.cgi?id=89431 which also had the
patch.

Cc: stable@vger.kernel.org
Fixes: b8fff407a180 ("mac80211: fix use-after-free in defragmentation")
Signed-off-by: Andreas Müller <goo@stapelspeicher.org>
[rewrite commit message]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: avoid using uninitialized stack data
Jes Sorensen [Wed, 10 Dec 2014 19:14:07 +0000 (14:14 -0500)]
mac80211: avoid using uninitialized stack data

Avoid a case where we would access uninitialized stack data if the AP
advertises HT support without 40MHz channel support.

Cc: stable@vger.kernel.org
Fixes: f3000e1b43f1 ("mac80211: fix broken use of VHT/20Mhz with some APs")
Signed-off-by: Jes Sorensen <Jes.Sorensen@redhat.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agor8169:update rtl8168g pcie ephy parameter
Chun-Hao Lin [Wed, 10 Dec 2014 13:28:38 +0000 (21:28 +0800)]
r8169:update rtl8168g pcie ephy parameter

Add ephy parameter to rtl8168g.
Also change the common function of rtl8168g from "rtl_hw_start_8168g_1" to
 "rtl_hw_start_8168g". And function "rtl_hw_start_8168g_1" is used for
setting rtl8168g hardware parameters.

Following is the explanation of what hardware parameter change for.
rtl8168g may erroneous judge the PCIe signal quality and show the error bit
on PCI configuration space when in PCIe low power mode.
The following ephy parameters are for above issue.
{ 0x00, 0x0000, 0x0008 }
{ 0x0c, 0x37d0, 0x0820 }
{ 0x1e, 0x0000, 0x0001 }

rtl8168g may return to PCIe L0 from PCIe L0s low power mode too slow.
The following ephy parameter is for above issue.
{ 0x19, 0x8000, 0x0000 }

Signed-off-by: Chunhao Lin <hau@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: dsa: bcm_sf2: force link for all fixed PHY devices
Florian Fainelli [Fri, 12 Dec 2014 02:12:42 +0000 (18:12 -0800)]
net: dsa: bcm_sf2: force link for all fixed PHY devices

For ports of the switch that we define as "fixed PHYs" such as MoCA, we
would have our Port 7 special handling that would allow us to assert the
link status indication.

For other ports, such as e.g: RGMII_1 connected to a cable modem, we
would rely on whatever the bootloader has left configured, which is a
bad assumption to make, we really need to force the link status
indication here.

Fixes: 246d7f773c13 ("net: dsa: add Broadcom SF2 switch driver")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'dma_mb'
David S. Miller [Fri, 12 Dec 2014 02:15:37 +0000 (21:15 -0500)]
Merge branch 'dma_mb'

Alexander Duyck says:

====================
arch: Add lightweight memory barriers for coherent memory access

These patches introduce two new primitives for synchronizing cache coherent
memory writes and reads.  These two new primitives are:

dma_rmb()
dma_wmb()

The first patch cleans up some unnecessary overhead related to the
definition of read_barrier_depends, smp_read_barrier_depends, and comments
related to the barrier.

The second patch adds the primitives for the applicable architectures and
asm-generic.

The third patch adds the barriers to r8169 which turns out to be a good
example of where the new barriers might be useful as they have full
rmb()/wmb() barriers ordering accesses to the descriptors and the DescOwn
bit.

The fourth patch adds support for coherent_rmb() to the Intel fm10k, igb,
and ixgbe drivers.  Testing with the ixgbe driver has shown a processing
time reduction of at least 7ns per 64B frame on a Core i7-4930K.

This patch series is essentially the v7 for:
v4-7: Add lightweight memory barriers for coherent memory access
v3: Add lightweight memory barriers fast_rmb() and fast_wmb()
v2: Introduce load_acquire() and store_release()
v1: Introduce read_acquire()

The key changes in this patch series versus the earlier patches are:
v7 resubmit:
- Added Acked-by: Ben Herrenschmidt from v5 to dma_rmb/wmb patch
- No code changes from previous set, still applies cleanly and builds.
v7:
- Dropped test/debug patch that was accidentally slipped in
v6:
- Replaced "memory based device I/O" with "consistent memory" in
  docs
- Added reference to DMA-API.txt to explain consistent memory
v5:
- Renamed barriers dma_rmb and dma_wmb
- Undid smp_wmb changes in x86 and PowerPC
- Defined smp_rmb as __lwsync for SMP case on PowerPC
v4:
- Renamed barriers coherent_rmb and coherent_wmb
- Added smp_lwsync for use in smp_load_acquire/smp_store_release
v3:
- Moved away from acquire()/store() and instead focused on barriers
- Added cleanup of read_barrier_depends
- Added change in r8169 to fix cur_tx/DescOwn ordering
- Simplified changes to just replacing/moving barriers in r8169
- Added update to documentation with code example
v2:
- Renamed read_acquire() to be consistent with smp_load_acquire()
- Changed barrier used to be consistent with smp_load_acquire()
- Updated PowerPC code to use __lwsync based on IBM article
- Added store_release() as this is a viable use case for drivers
- Added r8169 patch which is able to fully use primitives
- Added fm10k/igb/ixgbe patch which is able to test performance
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agofm10k/igb/ixgbe: Use dma_rmb on Rx descriptor reads
Alexander Duyck [Thu, 11 Dec 2014 23:02:28 +0000 (15:02 -0800)]
fm10k/igb/ixgbe: Use dma_rmb on Rx descriptor reads

This change makes it so that dma_rmb is used when reading the Rx
descriptor.  The advantage of dma_rmb is that it allows for a much
lower cost barrier on x86, powerpc, arm, and arm64 architectures than a
traditional memory barrier when dealing with reads that only have to
synchronize to coherent memory.

In addition I have updated the code so that it just checks to see if any
bits have been set instead of just the DD bit since the DD bit will always
be set as a part of a descriptor write-back so we just need to check for a
non-zero value being present at that memory location rather than just
checking for any specific bit.  This allows the code itself to appear much
cleaner and allows the compiler more room to optimize.

Cc: Matthew Vick <matthew.vick@intel.com>
Cc: Don Skidmore <donald.c.skidmore@intel.com>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agor8169: Use dma_rmb() and dma_wmb() for DescOwn checks
Alexander Duyck [Thu, 11 Dec 2014 23:02:17 +0000 (15:02 -0800)]
r8169: Use dma_rmb() and dma_wmb() for DescOwn checks

The r8169 use a pair of wmb() calls when setting up the descriptor rings.
The first is to synchronize the descriptor data with the descriptor status,
and the second is to synchronize the descriptor status with the use of the
MMIO doorbell to notify the device that descriptors are ready.  This can
come at a heavy price on some systems, and is not really necessary on
systems such as x86 as a simple barrier() would suffice to order store/store
accesses.  As such we can replace the first memory barrier with
dma_wmb() to reduce the cost for these accesses.

In addition the r8169 uses a rmb() to prevent compiler optimization in the
cleanup paths, however by moving the barrier down a few lines and replacing
it with a dma_rmb() we should be able to use it to guarantee
descriptor accesses do not occur until the device has updated the DescOwn
bit from its end.

One last change I made is to move the update of cur_tx in the xmit path to
after the wmb.  This way we can guarantee the device and all CPUs should
see the DescOwn update before they see the cur_tx value update.

Cc: Realtek linux nic maintainers <nic_swsd@realtek.com>
Cc: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoarch: Add lightweight memory barriers dma_rmb() and dma_wmb()
Alexander Duyck [Thu, 11 Dec 2014 23:02:06 +0000 (15:02 -0800)]
arch: Add lightweight memory barriers dma_rmb() and dma_wmb()

There are a number of situations where the mandatory barriers rmb() and
wmb() are used to order memory/memory operations in the device drivers
and those barriers are much heavier than they actually need to be.  For
example in the case of PowerPC wmb() calls the heavy-weight sync
instruction when for coherent memory operations all that is really needed
is an lsync or eieio instruction.

This commit adds a coherent only version of the mandatory memory barriers
rmb() and wmb().  In most cases this should result in the barrier being the
same as the SMP barriers for the SMP case, however in some cases we use a
barrier that is somewhere in between rmb() and smp_rmb().  For example on
ARM the rmb barriers break down as follows:

  Barrier   Call     Explanation
  --------- -------- ----------------------------------
  rmb()     dsb()    Data synchronization barrier - system
  dma_rmb() dmb(osh) data memory barrier - outer sharable
  smp_rmb() dmb(ish) data memory barrier - inner sharable

These new barriers are not as safe as the standard rmb() and wmb().
Specifically they do not guarantee ordering between coherent and incoherent
memories.  The primary use case for these would be to enforce ordering of
reads and writes when accessing coherent memory that is shared between the
CPU and a device.

It may also be noted that there is no dma_mb().  Most architectures don't
provide a good mechanism for performing a coherent only full barrier without
resorting to the same mechanism used in mb().  As such there isn't much to
be gained in trying to define such a function.

Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Michael Ellerman <michael@ellerman.id.au>
Cc: Michael Neuling <mikey@neuling.org>
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Tony Luck <tony.luck@intel.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: David Miller <davem@davemloft.net>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoarch: Cleanup read_barrier_depends() and comments
Alexander Duyck [Thu, 11 Dec 2014 23:01:55 +0000 (15:01 -0800)]
arch: Cleanup read_barrier_depends() and comments

This patch is meant to cleanup the handling of read_barrier_depends and
smp_read_barrier_depends.  In multiple spots in the kernel headers
read_barrier_depends is defined as "do {} while (0)", however we then go
into the SMP vs non-SMP sections and have the SMP version reference
read_barrier_depends, and the non-SMP define it as yet another empty
do/while.

With this commit I went through and cleaned out the duplicate definitions
and reduced the number of definitions down to 2 per header.  In addition I
moved the 50 line comments for the macro from the x86 and mips headers that
defined it as an empty do/while to those that were actually defining the
macro, alpha and blackfin.

Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge branch 'dsa'
David S. Miller [Fri, 12 Dec 2014 02:00:13 +0000 (21:00 -0500)]
Merge branch 'dsa'

Florian Fainelli says:

====================
net: dsa: two small bug fixes

Here are two small fixes for the DSA slave interface creation code:

- first patch fixes a null pointer de-reference with an invalid PHY
  device pointer while calling phy_connect_direct()

- second path propagates the dsa_slave_phy_setup() error code down to
  its caller: dsa_slave_create
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: dsa: propagate error code from dsa_slave_phy_setup
Florian Fainelli [Thu, 11 Dec 2014 20:49:16 +0000 (12:49 -0800)]
net: dsa: propagate error code from dsa_slave_phy_setup

In case we cannot attach to our slave netdevice PHY, error out and
propagate that error up to the caller: dsa_slave_create().

Fixes: 0d8bcdd383b8 ("net: dsa: allow for more complex PHY setups")
Signed-off-by: Andrey Volkov <andrey.volkov@nexvision.fr>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet: dsa: handle non-existing PHYs on switch internal bus
Florian Fainelli [Thu, 11 Dec 2014 20:49:15 +0000 (12:49 -0800)]
net: dsa: handle non-existing PHYs on switch internal bus

In case there is no PHY at the designated address on the internal
switch, we would basically de-reference a null pointer here:

dsa_slave_phy_setup(...)
{
p->phy = ds->slave_mii_bus->phy_map[p->port];
phy_connect_direct(slave_dev, p->phy, dsa_slave_adjust_link,
      ^------

This can be triggered when the platform configuration (platform_data or
Device Tree) indicates there should be a PHY device at this address, but
the HW is non-responsive, such that we cannot attach a PHY device at
this specific location.

Fix this by checking the return value prior to calling
phy_connect_direct().

CC: Andrew Lunn <andrew@lunn.ch>
Fixes: b31f65fb4383 ("net: dsa: slave: Fix autoneg for phys on switch MDIO bus")
Reported-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Andrey Volkov <andrey.volkov@nexvision.fr>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next
Linus Torvalds [Thu, 11 Dec 2014 22:27:06 +0000 (14:27 -0800)]
Merge git://git./linux/kernel/git/davem/net-next

Pull networking updates from David Miller:

 1) New offloading infrastructure and example 'rocker' driver for
    offloading of switching and routing to hardware.

    This work was done by a large group of dedicated individuals, not
    limited to: Scott Feldman, Jiri Pirko, Thomas Graf, John Fastabend,
    Jamal Hadi Salim, Andy Gospodarek, Florian Fainelli, Roopa Prabhu

 2) Start making the networking operate on IOV iterators instead of
    modifying iov objects in-situ during transfers.  Thanks to Al Viro
    and Herbert Xu.

 3) A set of new netlink interfaces for the TIPC stack, from Richard
    Alpe.

 4) Remove unnecessary looping during ipv6 routing lookups, from Martin
    KaFai Lau.

 5) Add PAUSE frame generation support to gianfar driver, from Matei
    Pavaluca.

 6) Allow for larger reordering levels in TCP, which are easily
    achievable in the real world right now, from Eric Dumazet.

 7) Add a variable of napi_schedule that doesn't need to disable cpu
    interrupts, from Eric Dumazet.

 8) Use a doubly linked list to optimize neigh_parms_release(), from
    Nicolas Dichtel.

 9) Various enhancements to the kernel BPF verifier, and allow eBPF
    programs to actually be attached to sockets.  From Alexei
    Starovoitov.

10) Support TSO/LSO in sunvnet driver, from David L Stevens.

11) Allow controlling ECN usage via routing metrics, from Florian
    Westphal.

12) Remote checksum offload, from Tom Herbert.

13) Add split-header receive, BQL, and xmit_more support to amd-xgbe
    driver, from Thomas Lendacky.

14) Add MPLS support to openvswitch, from Simon Horman.

15) Support wildcard tunnel endpoints in ipv6 tunnels, from Steffen
    Klassert.

16) Do gro flushes on a per-device basis using a timer, from Eric
    Dumazet.  This tries to resolve the conflicting goals between the
    desired handling of bulk vs.  RPC-like traffic.

17) Allow userspace to ask for the CPU upon what a packet was
    received/steered, via SO_INCOMING_CPU.  From Eric Dumazet.

18) Limit GSO packets to half the current congestion window, from Eric
    Dumazet.

19) Add a generic helper so that all drivers set their RSS keys in a
    consistent way, from Eric Dumazet.

20) Add xmit_more support to enic driver, from Govindarajulu
    Varadarajan.

21) Add VLAN packet scheduler action, from Jiri Pirko.

22) Support configurable RSS hash functions via ethtool, from Eyal
    Perry.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1820 commits)
  Fix race condition between vxlan_sock_add and vxlan_sock_release
  net/macb: fix compilation warning for print_hex_dump() called with skb->mac_header
  net/mlx4: Add support for A0 steering
  net/mlx4: Refactor QUERY_PORT
  net/mlx4_core: Add explicit error message when rule doesn't meet configuration
  net/mlx4: Add A0 hybrid steering
  net/mlx4: Add mlx4_bitmap zone allocator
  net/mlx4: Add a check if there are too many reserved QPs
  net/mlx4: Change QP allocation scheme
  net/mlx4_core: Use tasklet for user-space CQ completion events
  net/mlx4_core: Mask out host side virtualization features for guests
  net/mlx4_en: Set csum level for encapsulated packets
  be2net: Export tunnel offloads only when a VxLAN tunnel is created
  gianfar: Fix dma check map error when DMA_API_DEBUG is enabled
  cxgb4/csiostor: Don't use MASTER_MUST for fw_hello call
  net: fec: only enable mdio interrupt before phy device link up
  net: fec: clear all interrupt events to support i.MX6SX
  net: fec: reset fep link status in suspend function
  net: sock: fix access via invalid file descriptor
  net: introduce helper macro for_each_cmsghdr
  ...

9 years agoMerge tag 'sound-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Thu, 11 Dec 2014 21:20:50 +0000 (13:20 -0800)]
Merge tag 'sound-3.19-rc1' of git://git./linux/kernel/git/tiwai/sound

Pull sound updates from Takashi Iwai:
 "This became a fairly large pull request.  In addition to the usual
  driver updates / fixes, there have been a high amount of cleanups in
  ASoC area, as well as control API helpers and kernel documentations
  fixes touching through the whole tree.

  In the driver side, the biggest changes are the support for new Intel
  SoC found on new x86 machines, and the updates of FireWire dice and
  oxfw drivers.

  Some remarkable items are below:

  ALSA core:
   - PCM mmap code cleanup, removal of arch-dependent codes
   - PCM xrun injection support
   - PCM hwptr tracepoint support
   - Refactoring of snd_pcm_action(), simplification of PCM locking
   - Robustified sequecner auto-load functionality
   - New control API helpers and lots of cleanups along with them
   - Lots of kerneldoc fixes and cleanups

  USB-audio:
   - The mixer resume code was largely rewritten, and the devices with
     quirks are resumed properly.
   - New hardware support: Focusrite Scarlett, Digidesign Mbox1,
     Denon/Marantz DACs, Zoom R16/24

  FireWire:
   - DICE driver updates with better duplex and sync support, including
     MIDI support
   - New OXFW driver for Oxford Semiconductor FW970/971 chipset,
     including the previous LaCie Speakers device.  Fullduplex and MIDI
     support included as well as DICE driver.

  HD-audio:
   - Refactoring the driver-caps quirk handling in snd-hda-intel
   - More consistent control names representing the topology better
   - Fixups: HP mute LED with ALC268 codec, Ideapad S210 built-in mic
     fix, ASUS Z99He laptop EAPD

  ASoC:
   - Conversion of AC'97 drivers to use regmap, bringing us closer to
     the removal of the ASoC level I/O code
   - Clean up a lot of old drivers that were open coding things that
     have subsequently been implemented in the core
   - Some DAPM performance improvements
   - Removal of the now seldom used CODEC mutex
   - Lots of updates for the newer Intel SoC support, including support
     for the DSP and some Cherrytrail and Braswell machine drivers
   - Support for Samsung boards using rt5631 as the CODEC
   - Removal of the obsolete AFEB9260 machine driver
   - Driver support for the TI TS3A227E headset driver used in some
     Chrombeooks

  Others:
   - ASIHPI driver update and cleanups
   - Lots of dev_*() printk conversions
   - Lots of trivial cleanups for the codes spotted by Coccinelle"

* tag 'sound-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (594 commits)
  ALSA: pcxhr: NULL dereference on probe failure
  ALSA: lola: NULL dereference on probe failure
  ALSA: hda - Add "eapd" model string for AD1986A codec
  ALSA: hda - Add EAPD fixup for ASUS Z99He laptop
  ALSA: oxfw: Add hwdep interface
  ALSA: oxfw: Add support for capture/playback MIDI messages
  ALSA: oxfw: add support for capturing PCM samples
  ALSA: oxfw: Add support AMDTP in-stream
  ALSA: oxfw: Add support for Behringer/Mackie devices
  ALSA: oxfw: Change the way to start stream
  ALSA: oxfw: Add proc interface for debugging purpose
  ALSA: oxfw: Change the way to make PCM rules/constraints
  ALSA: oxfw: Add support for AV/C stream format command to get/set supported stream formation
  ALSA: oxfw: Change the way to name card
  ALSA: dice: Add support for MIDI capture/playback
  ALSA: dice: Add support for capturing PCM samples
  ALSA: dice: Support for non SYT-Match sampling clock source mode
  ALSA: dice: Add support for duplex streams with synchronization
  ALSA: dice: Change the way to start stream
  ALSA: jack: Add dummy snd_jack_set_key() definition
  ...

9 years agoMerge tag 'devicetree-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Thu, 11 Dec 2014 21:06:58 +0000 (13:06 -0800)]
Merge tag 'devicetree-for-linus' of git://git./linux/kernel/git/glikely/linux

Pull devicetree changes from Grant Likely:
 "Lots of activity in the devicetree code for v3.18.  Most of it is
  related to getting all of the overlay support code in place, but there
  are other important things in there.

  Highlights:

   - OF_RECONFIG notifiers for SPI, I2C and Platform devices.  Those
     subsystems can now respond to live changes to the device tree.

   - CONFIG_OF_OVERLAY method for applying live changes to the device
     tree

   - Removal of the of_allnodes list.  This used to be used to iterate
     over all the nodes in the device tree, but it is unnecessary
     because the same thing can be done by iterating over the list of
     child pointers.  Getting rid of of_allnodes saves some memory and
     avoids the possibility of of_allnodes being sorted differently from
     the child lists.

   - Support for retrieving original DTB blob via sysfs.  Needed by
     kexec.

   - More unittests

   - Documentation and minor bug fixes"

* tag 'devicetree-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/glikely/linux: (42 commits)
  of: Delete unnecessary check before calling "of_node_put()"
  of: Drop ->next pointer from struct device_node
  spi: Check for spi_of_notifier when CONFIG_OF_DYNAMIC=y
  of: support passing console options with stdout-path
  of: add optional options parameter to of_find_node_by_path()
  of: Add bindings for chosen node, stdout-path
  of: Remove unneeded and incorrect MODULE_DEVICE_TABLE
  ARM: dt: fix up PL011 device tree bindings
  of: base, fix of_property_read_string_helper kernel-doc
  of: remove select of non-existant OF_DEVICE config symbol
  spi/of: Add OF notifier handler
  spi/of: Create new device registration method and accessors
  i2c/of: Add OF_RECONFIG notifier handler
  i2c/of: Factor out Devicetree registration code
  of/overlay: Add overlay unittests
  of/overlay: Introduce DT overlay support
  of/reconfig: Add OF_DYNAMIC notifier for platform_bus_type
  of/reconfig: Always use the same structure for notifiers
  of/reconfig: Add debug output for OF_RECONFIG notifiers
  of/reconfig: Add empty stubs for the of_reconfig methods
  ...

9 years agoMerge tag 'fbdev-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux
Linus Torvalds [Thu, 11 Dec 2014 20:46:32 +0000 (12:46 -0800)]
Merge tag 'fbdev-3.19' of git://git./linux/kernel/git/tomba/linux

Pull fbdev updates from Tomi Valkeinen:
 - support for mx6sl and mx6sx
 - OMAP HDMI audio rewrite to make it finally work
 - OMAP video PLL work to prepare for new DRA7xx SoCs
 - simplefb DT related improvements

* tag 'fbdev-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux: (81 commits)
  video: uvesafb: Deletion of an unnecessary check before the function call "platform_device_put"
  video: fbdev-VIA: Deletion of an unnecessary check before the function call "framebuffer_release"
  video: fbdev-MMP: Deletion of an unnecessary check before the function call "mmp_unregister_path"
  video: mx3fb: Deletion of an unnecessary check before the function call "backlight_device_unregister"
  video: fbdev-OMAP2: Deletion of unnecessary checks before the function call "i2c_put_adapter"
  video: fbdev-SIS: Deletion of unnecessary checks before the function call "pci_dev_put"
  video: smscufx: Deletion of unnecessary checks before the function call "vfree"
  video: udlfb: Deletion of unnecessary checks before the function call "vfree"
  video: uvesafb: Deletion of an unnecessary check before the function call "uvesafb_free"
  video: fbdev-LCDC: Deletion of an unnecessary check before the function call "vfree"
  video: fbdev: arkfb: suppress build warning
  video: fbdev: s3fb: suppress build warning
  video: fbdev: vt8623fb: suppress build warning
  OMAPDSS: hdmi5: Fix bit field for IEC958_AES2_CON_SOURCE
  OMAPDSS: hdmi: Remove __exit qualifier from hdmi_uninit_output()
  OMAPDSS: hdmi5: Change hdmi_wp idlemode to to no_idle for audio playback
  OMAPDSS: Remove all references to obsolete HDMI audio callbacks
  ASoC: omap: Remove obsolete HDMI audio code and Kconfig options
  OMAPDSS: hdmi5: Register ASoC platform device for omap hdmi audio
  OMAPDSS: hdmi5: Remove callbacks for the old ASoC DAI driver
  ...

9 years agoMerge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Linus Torvalds [Thu, 11 Dec 2014 20:20:31 +0000 (12:20 -0800)]
Merge tag 'for_linus' of git://git./linux/kernel/git/mst/vhost

Pull virtio updates from Michael Tsirkin:
 "virtio: virtio 1.0 support, misc patches

  This adds a lot of infrastructure for virtio 1.0 support.  Notable
  missing pieces: virtio pci, virtio balloon (needs spec extension),
  vhost scsi.

  Plus, there are some minor fixes in a couple of places.

  Note: some net drivers are affected by these patches.  David said he's
  fine with merging these patches through my tree.

  Rusty's on vacation, he acked using my tree for these, too"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost: (70 commits)
  virtio_ccw: finalize_features error handling
  virtio_ccw: future-proof finalize_features
  virtio_pci: rename virtio_pci -> virtio_pci_common
  virtio_pci: update file descriptions and copyright
  virtio_pci: split out legacy device support
  virtio_pci: setup config vector indirectly
  virtio_pci: setup vqs indirectly
  virtio_pci: delete vqs indirectly
  virtio_pci: use priv for vq notification
  virtio_pci: free up vq->priv
  virtio_pci: fix coding style for structs
  virtio_pci: add isr field
  virtio: drop legacy_only driver flag
  virtio_balloon: drop legacy_only driver flag
  virtio_ccw: rev 1 devices set VIRTIO_F_VERSION_1
  virtio: allow finalize_features to fail
  virtio_ccw: legacy: don't negotiate rev 1/features
  virtio: add API to detect legacy devices
  virtio_console: fix sparse warnings
  vhost: remove unnecessary forward declarations in vhost.h
  ...

9 years agoMerge branch 'mailbox-devel' of git://git.linaro.org/landing-teams/working/fujitsu...
Linus Torvalds [Thu, 11 Dec 2014 20:09:37 +0000 (12:09 -0800)]
Merge branch 'mailbox-devel' of git://git.linaro.org/landing-teams/working/fujitsu/integration

Pull mailbox framework updates from Jassi Brar.

* 'mailbox-devel' of git://git.linaro.org/landing-teams/working/fujitsu/integration:
  Mailbox: Add support for Platform Communication Channel
  mailbox/omap: adapt to the new mailbox framework
  mailbox: add tx_prepare client callback
  mailbox: Don't unnecessarily re-arm the polling timer

9 years agoMerge tag 'spi-v3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Linus Torvalds [Thu, 11 Dec 2014 20:03:34 +0000 (12:03 -0800)]
Merge tag 'spi-v3.19' of git://git./linux/kernel/git/broonie/spi

Pull spi updates from Mark Brown:
 "Not a huge amount going on this release, mainly new drivers (there's a
  couple more waiting that didn't quite make the cut for this release
  too):

   - An interface for querying if the current transfer is the last in a
     message, allowing controllers that need special handling for the
     final transfer to use the core message parsing.
   - Support for Amlogic Meson SPIFC, Imagination Technologies SFPI,
     Intel Quark X1000 and Samsung Exynos 7 controllers"

* tag 'spi-v3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi: (38 commits)
  spi/s3c64xx: Remove redundant runtime PM management
  spi: fsl-spi: remove unused variable assignment
  spi: spi-fsl-spi: Return an error code in fsl_spi_do_one_msg()
  spi: core: Do not mangle error code from kthread_run()
  spi: fsl-espi: add (un)prepare_transfer_hardware calls to save power if SPI is not in use
  spi: fsl-(e)spi: migrate to generic master queueing
  spi/txx9: Deletion of an unnecessary check before the function call "clk_disable"
  spi: cadence: Fix 3-to-8 mux mode
  spi: cadence: Init HW after reading devicetree attributes
  spi: meson: Select REGMAP_MMIO
  spi: s3c64xx: add support for exynos7 SPI controller
  spi: spi-pxa2xx: SPI support for Intel Quark X1000
  spi: meson: meson_spifc_setup_speed() can be static
  spi: spi-pxa2xx: Add helpers for regiseters' accessing
  spi: spi-mxs: Fix mapping from vmalloc-ed buffer to scatter list
  spi: atmel: introduce probe deferring
  spi: atmel: remove compat for non DT board when requesting dma chan
  spi: meson: Add support for Amlogic Meson SPIFC
  spi: meson: Add device tree bindings documentation for SPIFC
  spi: core: Add spi_transfer_is_last() helper
  ...

9 years agoMerge tag 'edac/v3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Thu, 11 Dec 2014 19:58:50 +0000 (11:58 -0800)]
Merge tag 'edac/v3.19-rc1' of git://git./linux/kernel/git/mchehab/linux-edac

Pull edac updates from Mauro Carvalho Chehab:
 - Broadwell-DE support on sb-edac driver
 - Some fixes at sb-edac driver

* tag 'edac/v3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edac:
  sb_edac: Fix typo computing number of banks
  sb_edac: Add support for Broadwell-DE processor
  sb_edac: Fix discovery of top-of-low-memory for Haswell
  sb_edac: Fix erroneous bytes->gigabytes conversion
  sb_edac: Fix off-by-one error in number of channels

9 years agoFix race condition between vxlan_sock_add and vxlan_sock_release
Marcelo Leitner [Thu, 11 Dec 2014 12:02:22 +0000 (10:02 -0200)]
Fix race condition between vxlan_sock_add and vxlan_sock_release

Currently, when trying to reuse a socket, vxlan_sock_add will grab
vn->sock_lock, locate a reusable socket, inc refcount and release
vn->sock_lock.

But vxlan_sock_release() will first decrement refcount, and then grab
that lock. refcnt operations are atomic but as currently we have
deferred works which hold vs->refcnt each, this might happen, leading to
a use after free (specially after vxlan_igmp_leave):

  CPU 1                            CPU 2

deferred work                    vxlan_sock_add
  ...                              ...
                                   spin_lock(&vn->sock_lock)
                                   vs = vxlan_find_sock();
  vxlan_sock_release
    dec vs->refcnt, reaches 0
    spin_lock(&vn->sock_lock)
                                   vxlan_sock_hold(vs), refcnt=1
                                   spin_unlock(&vn->sock_lock)
    hlist_del_rcu(&vs->hlist);
    vxlan_notify_del_rx_port(vs)
    spin_unlock(&vn->sock_lock)

So when we look for a reusable socket, we check if it wasn't freed
already before reusing it.

Signed-off-by: Marcelo Ricardo Leitner <mleitner@redhat.com>
Fixes: 7c47cedf43a8b3 ("vxlan: move IGMP join/leave to work queue")
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/macb: fix compilation warning for print_hex_dump() called with skb->mac_header
Cyrille Pitchen [Thu, 11 Dec 2014 10:15:54 +0000 (11:15 +0100)]
net/macb: fix compilation warning for print_hex_dump() called with skb->mac_header

Signed-off-by: Cyrille Pitchen <cyrille.pitchen@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge tag 'media/v3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Thu, 11 Dec 2014 19:49:23 +0000 (11:49 -0800)]
Merge tag 'media/v3.19-rc1' of git://git./linux/kernel/git/mchehab/linux-media

Pull media updates from Mauro Carvalho Chehab:
 - Two new dvb frontend drivers: mn88472 and mn88473
 - A new driver for some PCIe DVBSky cards
 - A new remote controller driver: meson-ir
 - One LIRC staging driver got rewritten and promoted to mainstream:
   igorplugusb
 - A new tuner driver (m88rs6000t)
 - The old omap2 media driver got removed from staging.  This driver
   uses an old DMA API and it is likely broken on recent kernels.
   Nobody cared enough to fix it
 - Media bus format moved to a separate header, as DRM will also use the
   definitions there
 - mem2mem_testdev were renamed to vim2m, in order to use the same
   naming convention taken by the other virtual test driver (vivid)
 - Added a new driver for coda SoC (coda-jpeg)
 - The cx88 driver got converted to use videobuf2 core
 - Make DMABUF export buffer to work with DMA Scatter/Gather and Vmalloc
   cores
 - Lots of other fixes, improvements and cleanups on the drivers.

* tag 'media/v3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media: (384 commits)
  [media] mn88473: One function call less in mn88473_init() after error
  [media] mn88473: Remove uneeded check before release_firmware()
  [media] lirc_zilog: Deletion of unnecessary checks before vfree()
  [media] MAINTAINERS: Add myself as img-ir maintainer
  [media] img-ir: Don't set driver's module owner
  [media] img-ir: Depend on METAG or MIPS or COMPILE_TEST
  [media] img-ir/hw: Drop [un]register_decoder declarations
  [media] img-ir/hw: Fix potential deadlock stopping timer
  [media] img-ir/hw: Always read data to clear buffer
  [media] redrat3: ensure dma is setup properly
  [media] ddbridge: remove unneeded check before dvb_unregister_device()
  [media] si2157: One function call less in si2157_init() after error
  [media] tuners: remove uneeded checks before release_firmware()
  [media] arm: omap2: rx51-peripherals: fix build warning
  [media] stv090x: add an extra protetion against buffer overflow
  [media] stv090x: Remove an unreachable code
  [media] stv090x: Some whitespace cleanups
  [media] em28xx: checkpatch cleanup: whitespaces/new lines cleanups
  [media] si2168: add support for firmware files in new format
  [media] si2168: debug printout for firmware version
  ...

9 years agoMerge branch 'mlx4-next'
David S. Miller [Thu, 11 Dec 2014 19:47:40 +0000 (14:47 -0500)]
Merge branch 'mlx4-next'

Or Gerlitz says:

====================
mlx4 driver update

This series from Matan, Jenny, Dotan and myself is mostly about adding
support to a new performance optimized flow steering mode (patches 4-10).

The 1st two patches are small fixes (one for VXLAN and one for SRIOV),
and the third patch is a fix to avoid hard-lockup situation when many
(hunderds) processes holding user-space QPs/CQs get events.

Matan and Or.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4: Add support for A0 steering
Matan Barak [Thu, 11 Dec 2014 08:58:00 +0000 (10:58 +0200)]
net/mlx4: Add support for A0 steering

Add the required firmware commands for A0 steering and a way to enable
that. The firmware support focuses on INIT_HCA, QUERY_HCA, QUERY_PORT,
QUERY_DEV_CAP and QUERY_FUNC_CAP commands. Those commands are used
to configure and query the device.

The different A0 DMFS (steering) modes are:

Static - optimized performance, but flow steering rules are
limited. This mode should be choosed explicitly by the user
in order to be used.

Dynamic - this mode should be explicitly choosed by the user.
In this mode, the FW works in optimized steering mode as long as
it can and afterwards automatically drops to classic (full) DMFS.

Disable - this mode should be explicitly choosed by the user.
The user instructs the system not to use optimized steering, even if
the FW supports Dynamic A0 DMFS (and thus will be able to use optimized
steering in Default A0 DMFS mode).

Default - this mode is implicitly choosed. In this mode, if the FW
supports Dynamic A0 DMFS, it'll work in this mode. Otherwise, it'll
work at Disable A0 DMFS mode.

Under SRIOV configuration, when the A0 steering mode is enabled,
older guest VF drivers who aren't using the RX QP allocation flag
(MLX4_RESERVE_A0_QP) will get a QP from the general range and
fail when attempting to register a steering rule. To avoid that,
the PF context behaviour is changed once on A0 static mode, to
require support for the allocation flag in VF drivers too.

In order to enable A0 steering, we use log_num_mgm_entry_size param.
If the value of the parameter is not positive, we treat the absolute
value of log_num_mgm_entry_size as a bit field. Setting bit 2 of this
bit field enables static A0 steering.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4: Refactor QUERY_PORT
Matan Barak [Thu, 11 Dec 2014 08:57:59 +0000 (10:57 +0200)]
net/mlx4: Refactor QUERY_PORT

Currently QUERY_PORT is done as a part of QUERY_DEV_CAP firmware command.

Since we would like to use it without querying all device capabilities,
extract this part to be a function of its own.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4_core: Add explicit error message when rule doesn't meet configuration
Matan Barak [Thu, 11 Dec 2014 08:57:58 +0000 (10:57 +0200)]
net/mlx4_core: Add explicit error message when rule doesn't meet configuration

When a given flow steering rule is invalid in respect to the current
steering configuration, print the correct error message to the system log.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4: Add A0 hybrid steering
Matan Barak [Thu, 11 Dec 2014 08:57:57 +0000 (10:57 +0200)]
net/mlx4: Add A0 hybrid steering

A0 hybrid steering is a form of high performance flow steering.
By using this mode, mlx4 cards use a fast limited table based steering,
in order to enable fast steering of unicast packets to a QP.

In order to implement A0 hybrid steering we allocate resources
from different zones:
(1) General range
(2) Special MAC-assigned QPs [RSS, Raw-Ethernet] each has its own region.

When we create a rss QP or a raw ethernet (A0 steerable and BF ready) QP,
we try hard to allocate the QP from range (2). Otherwise, we try hard not
to allocate from this  range. However, when the system is pushed to its
limits and one needs every resource, the allocator uses every region it can.

Meaning, when we run out of raw-eth qps, the allocator allocates from the
general range (and the special-A0 area is no longer active). If we run out
of RSS qps, the mechanism tries to allocate from the raw-eth QP zone. If that
is also exhausted, the allocator will allocate from the general range
(and the A0 region is no longer active).

Note that if a raw-eth qp is allocated from the general range, it attempts
to allocate the range such that bits 6 and 7 (blueflame bits) in the
QP number are not set.

When the feature is used in SRIOV, the VF has to notify the PF what
kind of QP attributes it needs. In order to do that, along with the
"Eth QP blueflame" bit, we reserve a new "A0 steerable QP". According
to the combination of these bits, the PF tries to allocate a suitable QP.

In order to maintain backward compatibility (with older PFs), the PF
notifies which QP attributes it supports via QUERY_FUNC_CAP command.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4: Add mlx4_bitmap zone allocator
Matan Barak [Thu, 11 Dec 2014 08:57:56 +0000 (10:57 +0200)]
net/mlx4: Add mlx4_bitmap zone allocator

The zone allocator is a mechanism which manages a few mlx4_bitmaps.

When allocating a resource, the user indicates the desired zone of
which this resource will be allocated from. If possible, the resource
will be allocated from this zone. Otherwise, the resource will be
allocated from a less-than, equal-to, higher-than priority zone,
according to the desired zone's properties with that respective
allocation order.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4: Add a check if there are too many reserved QPs
Dotan Barak [Thu, 11 Dec 2014 08:57:55 +0000 (10:57 +0200)]
net/mlx4: Add a check if there are too many reserved QPs

The number of reserved QPs is affected both from the firmware and
from the driver's requirements. This patch adds a check that
validates that this number is indeed feasable.

Signed-off-by: Dotan Barak <dotanb@dev.mellanox.co.il>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4: Change QP allocation scheme
Eugenia Emantayev [Thu, 11 Dec 2014 08:57:54 +0000 (10:57 +0200)]
net/mlx4: Change QP allocation scheme

When using BF (Blue-Flame), the QPN overrides the VLAN, CV, and SV fields
in the WQE. Thus, BF may only be used for QPNs with bits 6,7 unset.

The current Ethernet driver code reserves a Tx QP range with 256b alignment.

This is wrong because if there are more than 64 Tx QPs in use,
QPNs >= base + 65 will have bits 6/7 set.

This problem is not specific for the Ethernet driver, any entity that
tries to reserve more than 64 BF-enabled QPs should fail. Also, using
ranges is not necessary here and is wasteful.

The new mechanism introduced here will support reservation for
"Eth QPs eligible for BF" for all drivers: bare-metal, multi-PF, and VFs
(when hypervisors support WC in VMs). The flow we use is:

1. In mlx4_en, allocate Tx QPs one by one instead of a range allocation,
   and request "BF enabled QPs" if BF is supported for the function

2. In the ALLOC_RES FW command, change param1 to:
a. param1[23:0]  - number of QPs
b. param1[31-24] - flags controlling QPs reservation

Bit 31 refers to Eth blueflame supported QPs. Those QPs must have
bits 6 and 7 unset in order to be used in Ethernet.

Bits 24-30 of the flags are currently reserved.

When a function tries to allocate a QP, it states the required attributes
for this QP. Those attributes are considered "best-effort". If an attribute,
such as Ethernet BF enabled QP, is a must-have attribute, the function has
to check that attribute is supported before trying to do the allocation.

In a lower layer of the code, mlx4_qp_reserve_range masks out the bits
which are unsupported. If SRIOV is used, the PF validates those attributes
and masks out unsupported attributes as well. In order to notify VFs which
attributes are supported, the VF uses QUERY_FUNC_CAP command. This command's
mailbox is filled by the PF, which notifies which QP allocation attributes
it supports.

Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4_core: Use tasklet for user-space CQ completion events
Matan Barak [Thu, 11 Dec 2014 08:57:53 +0000 (10:57 +0200)]
net/mlx4_core: Use tasklet for user-space CQ completion events

Previously, we've fired all our completion callbacks straight from our ISR.

Some of those callbacks were lightweight (for example, mlx4_en's and
IPoIB napi callbacks), but some of them did more work (for example,
the user-space RDMA stack uverbs' completion handler). Besides that,
doing more than the minimal work in ISR is generally considered wrong,
it could even lead to a hard lockup of the system. Since when a lot
of completion events are generated by the hardware, the loop over those
events could be so long, that we'll get into a hard lockup by the system
watchdog.

In order to avoid that, add a new way of invoking completion events
callbacks. In the interrupt itself, we add the CQs which receive completion
event to a per-EQ list and schedule a tasklet. In the tasklet context
we loop over all the CQs in the list and invoke the user callback.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4_core: Mask out host side virtualization features for guests
Or Gerlitz [Thu, 11 Dec 2014 08:57:52 +0000 (10:57 +0200)]
net/mlx4_core: Mask out host side virtualization features for guests

When VFs (guests in this context) issue the QUERY_DEV_CAP command, they
need not be told that host side virtualization features such as VST, FSM
(MAC anti-spoofing) and running > 80 VFs are supported by the device.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4_en: Set csum level for encapsulated packets
Or Gerlitz [Thu, 11 Dec 2014 08:57:51 +0000 (10:57 +0200)]
net/mlx4_en: Set csum level for encapsulated packets

This was dropped by mistake for the napi_gro_frags flow, fix that.

Fixes: dd65beac48a5 ('net/mlx4_en: Extend usage of napi_gro_frags')
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoMerge tag 'backlight-for-linus-3.19' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Thu, 11 Dec 2014 19:39:03 +0000 (11:39 -0800)]
Merge tag 'backlight-for-linus-3.19' of git://git./linux/kernel/git/lee/backlight

Pull backlight updates from Lee Jones:
 - Clean-up leaky resources; pwm_bl
 - Simplify Device Tree initialisation; lp855x_bl
 - Add Regulator support; lp855x
 - Remove Bryan from the Maintainer list -- new baby, no time :)

* tag 'backlight-for-linus-3.19' of git://git.kernel.org/pub/scm/linux/kernel/git/lee/backlight:
  MAINTAINERS: Remove my name from Backlight subsystem
  backlight: lp855x: Add supply regulator to lp855x
  backlight: lp855x: Refactor DT parsing code
  backlight: pwm: Clean-up pwm requested using legacy API

9 years agobe2net: Export tunnel offloads only when a VxLAN tunnel is created
Sriharsha Basavapatna [Thu, 11 Dec 2014 08:24:47 +0000 (03:24 -0500)]
be2net: Export tunnel offloads only when a VxLAN tunnel is created

The encapsulated offload flags shouldn't be unconditionally exported
to the stack. The stack expects offloading to work across all tunnel
types when those flags are set. This would break other tunnels (like
GRE) since be2net currently supports tunnel offload for VxLAN only.

Also, with VxLANs Skyhawk-R can offload only 1 UDP dport. If more
than 1 UDP port is added, we should disable offloads in that case too.

Signed-off-by: Sriharsha Basavapatna <sriharsha.basavapatna@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agogianfar: Fix dma check map error when DMA_API_DEBUG is enabled
Kevin Hao [Thu, 11 Dec 2014 06:08:41 +0000 (14:08 +0800)]
gianfar: Fix dma check map error when DMA_API_DEBUG is enabled

We need to use dma_mapping_error() to check the dma address returned
by dma_map_single/page(). Otherwise we would get warning like this:
  WARNING: at lib/dma-debug.c:1140
  Modules linked in:
  CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.18.0-rc2-next-20141029 #196
  task: c0834300 ti: effe6000 task.ti: c0874000
  NIP: c02b2c98 LR: c02b2c98 CTR: c030abc4
  REGS: effe7d70 TRAP: 0700   Not tainted  (3.18.0-rc2-next-20141029)
  MSR: 00021000 <CE,ME>  CR: 22044022  XER: 20000000

  GPR00: c02b2c98 effe7e20 c0834300 00000098 00021000 00000000 c030b898 00000003
  GPR08: 00000001 00000000 00000001 749eec9d 22044022 1001abe0 00000020 ef278678
  GPR16: ef278670 ef278668 ef278660 070a8040 c087f99c c08cdc60 00029000 c0840d44
  GPR24: c08be6e8 c0840000 effe7e78 ef041340 00000600 ef114e10 00000000 c08be6e0
  NIP [c02b2c98] check_unmap+0x51c/0x9e4
  LR [c02b2c98] check_unmap+0x51c/0x9e4
  Call Trace:
  [effe7e20] [c02b2c98] check_unmap+0x51c/0x9e4 (unreliable)
  [effe7e70] [c02b31d8] debug_dma_unmap_page+0x78/0x8c
  [effe7ed0] [c03d1640] gfar_clean_rx_ring+0x208/0x488
  [effe7f40] [c03d1a9c] gfar_poll_rx_sq+0x3c/0xa8
  [effe7f60] [c04f8714] net_rx_action+0xc0/0x178
  [effe7f90] [c00435a0] __do_softirq+0x100/0x1fc
  [effe7fe0] [c0043958] irq_exit+0xa4/0xc8
  [effe7ff0] [c000d14c] call_do_irq+0x24/0x3c
  [c0875e90] [c00048a0] do_IRQ+0x8c/0xf8
  [c0875eb0] [c000ed10] ret_from_except+0x0/0x18

For TX, we need to unmap the pages which has already been mapped and
free the skb before return.

For RX, move the dma mapping and error check to gfar_new_skb(). We
would reuse the original skb in the rx ring when either allocating
skb failure or dma mapping error.

Signed-off-by: Kevin Hao <haokexin@gmail.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Reviewed-by: Claudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agocxgb4/csiostor: Don't use MASTER_MUST for fw_hello call
Hariprasad Shenai [Thu, 11 Dec 2014 05:41:43 +0000 (11:11 +0530)]
cxgb4/csiostor: Don't use MASTER_MUST for fw_hello call

Remove use of calls into t4_fw_hello() with MASTER_MUST, which results in
FW_HELLO_CMD_MASTERFORCE being set. The firmware doesn't support this and of
course any existing PF Drivers will totally go for a toss.

Signed-off-by: Hariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>