firefly-linux-kernel-4.4.55.git
12 years agosfc: Implement ethtool RX NFC rules API instead of n-tuple API
Ben Hutchings [Tue, 3 Jan 2012 12:05:47 +0000 (12:05 +0000)]
sfc: Implement ethtool RX NFC rules API instead of n-tuple API

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosfc: Add support for retrieving and removing filters by ID
Ben Hutchings [Tue, 3 Jan 2012 12:05:39 +0000 (12:05 +0000)]
sfc: Add support for retrieving and removing filters by ID

These new functions will support an implementation of the ethtool
RX NFC rules API.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosfc: Use consistent types for filter IDs, indices and search depths
Ben Hutchings [Tue, 3 Jan 2012 12:05:27 +0000 (12:05 +0000)]
sfc: Use consistent types for filter IDs, indices and search depths

Filter IDs are u32 (but never very large) so an ID/error return
value should have type s32.

Filter indices and search depths are never negative, so should
have type unsigned int.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosfc: Change filter ID generation to satisfy priority semantics of RX NFC
Ben Hutchings [Tue, 3 Jan 2012 12:05:15 +0000 (12:05 +0000)]
sfc: Change filter ID generation to satisfy priority semantics of RX NFC

Also add note that the efx_filter_spec::priority field has nothing
to do with priority between multiple matching filters.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoethtool: Allow drivers to select RX NFC rule locations
Ben Hutchings [Tue, 3 Jan 2012 12:04:51 +0000 (12:04 +0000)]
ethtool: Allow drivers to select RX NFC rule locations

Define special location values for RX NFC that request the driver to
select the actual rule location.  This allows for implementation on
devices that use hash-based filter lookup, whereas currently the API is
more suited to devices with TCAM lookup or linear search.

In ethtool_set_rxnfc() and the compat wrapper ethtool_ioctl(), copy
the structure back to user-space after insertion so that the actual
location is returned.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agogianfar: Reject out-of-range RX NFC locations
Ben Hutchings [Tue, 3 Jan 2012 11:59:30 +0000 (11:59 +0000)]
gianfar: Reject out-of-range RX NFC locations

Currently the driver only uses location values to maintain an ordered
list of filters.  Make it reject location values >= MAX_FILER_IDX
passed to the ETHTOOL_SRXCLSRLINS command, consistent with the range
it reports for the ETHTOOL_GRXCLSRLALL command.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Acked-by: Sebastian Pöhn <sebastian.poehn@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/smsc911x: Check if PHY is in operational mode before software reset
Javier Martinez Canillas [Tue, 3 Jan 2012 13:36:19 +0000 (13:36 +0000)]
net/smsc911x: Check if PHY is in operational mode before software reset

SMSC LAN generation 4 chips integrate an IEEE 802.3 ethernet physical layer.
The PHY driver for this integrated chip enable an energy detect power-down mode.
When the PHY is in a power-down mode, it prevents the MAC portion chip to be
software reseted.

That means that if we compile the kernel with the configuration option SMSC_PHY
enabled and try to bring the network interface up without an cable plug-ed the
PHY will be in a low power mode and the software reset will fail returning -EIO
to user-space:

root@igep00x0:~# ifconfig eth0 up
ifconfig: SIOCSIFFLAGS: Input/output error

This patch disable the energy detect power-down mode before trying to software
reset the LAN chip and re-enables after it was reseted successfully.

Signed-off-by: Javier Martinez Canillas <javier@dowhile0.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: phy: smsc: Move SMSC PHY constants to <linux/smscphy.h>
Javier Martinez Canillas [Wed, 4 Jan 2012 01:23:18 +0000 (20:23 -0500)]
net: phy: smsc: Move SMSC PHY constants to <linux/smscphy.h>

SMSC generation 4 LAN chips integrate an IEEE 802.3 ethernet physical layer.
The ethernet driver for this family of devices needs to access the SMSC PHY
registers and bit-fields.

So, this patch moves these constants to a place where it can be used for both
the PHY and LAN drivers.

Signed-off-by: Javier Martinez Canillas <javier@dowhile0.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Tue, 3 Jan 2012 20:16:34 +0000 (15:16 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next into for-davem

Conflicts:
drivers/net/wireless/b43/dma.c
drivers/net/wireless/brcm80211/brcmfmac/dhd_linux.c

12 years agoskge: fix warning when CONFIG_PM is defined but not CONFIG_PM_SLEEP
Daniel Halperin [Tue, 3 Jan 2012 18:53:16 +0000 (13:53 -0500)]
skge: fix warning when CONFIG_PM is defined but not CONFIG_PM_SLEEP

drivers/net/ethernet/marvell/skge.c:4046: warning: ‘skge_suspend’ defined but not used
drivers/net/ethernet/marvell/skge.c:4071: warning: ‘skge_resume’ defined but not used

Signed-off-by: Daniel Halperin <dhalperi@cs.washington.edu>
Cc: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/davinci: do not use all descriptors for tx packets
Sascha Hauer [Tue, 3 Jan 2012 05:27:47 +0000 (05:27 +0000)]
net/davinci: do not use all descriptors for tx packets

The driver uses a shared pool for both rx and tx descriptors.
During open it queues fixed number of 128 descriptors for receive
packets. For each received packet it tries to queue another
descriptor. If this fails the descriptor is lost for rx.
The driver has no limitation on tx descriptors to use, so it
can happen during a nmap / ping -f attack that the driver
allocates all descriptors for tx and looses all rx descriptors.
The driver stops working then.
To fix this limit the number of tx descriptors used to half of
the descriptors available, the rx path uses the other half.

Tested on a custom board using nmap / ping -f to the board from
two different hosts.

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet_sched: qdisc_alloc_handle() can be too slow
Eric Dumazet [Tue, 3 Jan 2012 00:00:11 +0000 (00:00 +0000)]
net_sched: qdisc_alloc_handle() can be too slow

When trying to allocate ~32768 qdiscs using autohandle mechanism, we can
fill the space managed by kernel (handles in [8000-FFFF]:0000 range)

But O(N^2) qdisc_alloc_handle() loops 0x10000 times instead of 0x8000

time tc add qdisc add dev eth0 parent 10:7fff pfifo limit 10
RTNETLINK answers: Cannot allocate memory
real    1m54.826s
user    0m0.000s
sys     0m0.004s

INFO: rcu_sched_state detected stall on CPU 0 (t=60000 jiffies)

Half number of loops, and add a cond_resched() call.
We hold rtnl at this point.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Dave Taht <dave.taht@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosch_qfq: accurate wsum handling
Eric Dumazet [Mon, 2 Jan 2012 11:47:50 +0000 (11:47 +0000)]
sch_qfq: accurate wsum handling

We can underestimate q->wsum in case of "tc class replace ... qfq"
and/or qdisc_create_dflt() error.

wsum is not really used in fast path, only at qfq qdisc/class setup,
to catch user error.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosch_sfq: dont put new flow at the end of flows
Eric Dumazet [Sun, 1 Jan 2012 18:33:31 +0000 (18:33 +0000)]
sch_sfq: dont put new flow at the end of flows

SFQ enqueue algo puts a new flow _behind_ all pre-existing flows in the
circular list. In fact this is probably an old SFQ implementation bug.

100 Mbits = ~8333 full frames per second, or ~8 frames per ms.

With 50 flows, it means your "new flow" will have to wait 50 packets
being sent before its own packet. Thats the ~6ms.

We certainly can change SFQ to give a priority advantage to new flows,
so that next dequeued packet is taken from a new flow, not an old one.

Reported-by: Dave Taht <dave.taht@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomlx4_core: Elaborating limitation on VF port options
Yevgeny Petrilin [Mon, 2 Jan 2012 04:07:43 +0000 (04:07 +0000)]
mlx4_core: Elaborating limitation on VF port options

Showing which capabilities are not passed to VF
when executing QUERY_PORT

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomlx4_core: fix mtt range deallocation
Marcel Apfelbaum [Mon, 2 Jan 2012 04:07:39 +0000 (04:07 +0000)]
mlx4_core: fix mtt range deallocation

The mtt range was allocated in mtt units but deallocated
in segments. Among the rest, this caused crash during hotplug removal

Reported-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Marcel Apfelbaum <marcela@mellanox.co.il>
Reviewed-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Tested-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobonding: fix error handling if slave is busy (v2)
stephen hemminger [Sat, 31 Dec 2011 13:26:46 +0000 (13:26 +0000)]
bonding: fix error handling if slave is busy (v2)

If slave device already has a receive handler registered, then the
error unwind of bonding device enslave function is broken.

The following will leave a pointer to freed memory in the slave
device list, causing a later kernel panic.
# modprobe dummy
# ip li add dummy0-1 link dummy0 type macvlan
# modprobe bonding
# echo +dummy0 >/sys/class/net/bond0/bonding/slaves

The fix is to detach the slave (which removes it from the list)
in the unwind path.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Reviewed-by: Nicolas de Pesloüan <nicolas.2p.debian@free.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years ago8139cp: properly config rx mode after resuming
Jason Wang [Fri, 30 Dec 2011 23:44:42 +0000 (23:44 +0000)]
8139cp: properly config rx mode after resuming

Rx mode should be reset after resming, so unconditionally updating rx
mode rather than conditionally updating based on the value we
remembered, otherwise unexpected value may be used by the nic after
resuming.

btw. I find and test this when debugging guest hibernation in qemu, as
I did not have a 8139cp card in hand, this patch is untested in a
physical 8139cp card, plase review it carefully.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years ago8139cp/8139too: do not read into reserved registers
Jason Wang [Fri, 30 Dec 2011 23:44:33 +0000 (23:44 +0000)]
8139cp/8139too: do not read into reserved registers

delay_eeprom() use long read for Cfg9346 register(offset 0x50) which may read
into the area of reserved register(offset 0x53). Use byte read instead.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoixgbe: add support for new 82599 device.
Don Skidmore [Sat, 10 Dec 2011 06:49:43 +0000 (06:49 +0000)]
ixgbe: add support for new 82599 device.

This device uses an already existing DevID but since it supports
WoL we need to add the Sub DevID.  It's support of WoL is limited
to the first port.

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: add support for new 82599 device id
Emil Tantilov [Fri, 4 Nov 2011 06:43:29 +0000 (06:43 +0000)]
ixgbe: add support for new 82599 device id

Support for new 82599 based quad port adapter.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: add write flush in ixgbe_clock_out_i2c_byte()
Emil Tantilov [Fri, 4 Nov 2011 06:43:23 +0000 (06:43 +0000)]
ixgbe: add write flush in ixgbe_clock_out_i2c_byte()

I2C access is timing critical. Always do a write flush after writing
to the I2CCTL register.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: fix typo's
Stephen Hemminger [Thu, 22 Dec 2011 16:34:52 +0000 (16:34 +0000)]
ixgbe: fix typo's

Saw typo in one message, so decided to run spell checker.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: fix incorrect PHY register reads
Emil Tantilov [Sat, 10 Dec 2011 08:21:47 +0000 (08:21 +0000)]
ixgbe: fix incorrect PHY register reads

Fix some register reads that had the opcode and register parameters swapped.
Also use define instead of a magic (0x3) number.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Add flow control advertising to ethtool setting.
Carolyn Wyborny [Fri, 2 Dec 2011 00:03:15 +0000 (00:03 +0000)]
igb: Add flow control advertising to ethtool setting.

Added pause flag for bi-directional flow control advertising to ethtool
settings.

Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbevf: Fix register defines to correctly handle complex expressions
Alexander Duyck [Thu, 8 Dec 2011 06:36:28 +0000 (06:36 +0000)]
ixgbevf: Fix register defines to correctly handle complex expressions

This patch is meant to address possible issues with the IXGBEVF register
defines generating incorrect values when given a complex expression for the
register offset.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Mon, 2 Jan 2012 23:56:49 +0000 (18:56 -0500)]
Merge git://git./linux/kernel/git/davem/net

12 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/padovan/blueto...
John W. Linville [Mon, 2 Jan 2012 21:43:54 +0000 (16:43 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/padovan/bluetooth-next

12 years agoMerge branch 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6
Linus Torvalds [Mon, 2 Jan 2012 20:34:03 +0000 (12:34 -0800)]
Merge branch 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6

* 'devicetree/merge' of git://git.secretlab.ca/git/linux-2.6:
  dt/device: Fix auxdata matching to handle entries without a name override

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Mon, 2 Jan 2012 03:36:08 +0000 (19:36 -0800)]
Merge git://git./linux/kernel/git/davem/net

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  netfilter: ctnetlink: fix timeout calculation
  ipvs: try also real server with port 0 in backup server
  skge: restore rx multicast filter on resume and after config changes
  mlx4_en: nullify cq->vector field when closing completion queue

12 years agonetfilter: nfnetlink_acct: fix nfnl_acct_get operation
Pablo Neira Ayuso [Fri, 30 Dec 2011 19:32:26 +0000 (20:32 +0100)]
netfilter: nfnetlink_acct: fix nfnl_acct_get operation

The get operation was not sending the message that was built to
user-space. This patch also includes the appropriate handling for
the return value of netlink_unicast().

Moreover, fix error codes on error (for example, for non-existing
entry was uncorrect).

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
12 years agoMerge branch 'fix/asoc' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Sat, 31 Dec 2011 19:55:06 +0000 (11:55 -0800)]
Merge branch 'fix/asoc' of git://git./linux/kernel/git/tiwai/sound

* 'fix/asoc' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ASoC: wm8776: add missing break in sample size switch

12 years agogspca: Fix falling back to lower isoc alt settings
Mauro Carvalho Chehab [Sat, 31 Dec 2011 13:32:03 +0000 (11:32 -0200)]
gspca: Fix falling back to lower isoc alt settings

The current gspca core code has a regression where it no longer properly
falls back to lower alt settings when there is not enough bandwidth.

This causes many iso based usb-1 cameras to not work when plugged into a
usb2 hub or a sandybridge chipset motherboard!

This patch fixes this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agofutex: Fix uninterruptible loop due to gate_area
Hugh Dickins [Sat, 31 Dec 2011 19:44:01 +0000 (11:44 -0800)]
futex: Fix uninterruptible loop due to gate_area

It was found (by Sasha) that if you use a futex located in the gate
area we get stuck in an uninterruptible infinite loop, much like the
ZERO_PAGE issue.

While looking at this problem, PeterZ realized you'll get into similar
trouble when hitting any install_special_pages() mapping.  And are there
still drivers setting up their own special mmaps without page->mapping,
and without special VM or pte flags to make get_user_pages fail?

In most cases, if page->mapping is NULL, we do not need to retry at all:
Linus points out that even /proc/sys/vm/drop_caches poses no problem,
because it ends up using remove_mapping(), which takes care not to
interfere when the page reference count is raised.

But there is still one case which does need a retry: if memory pressure
called shmem_writepage in between get_user_pages_fast dropping page
table lock and our acquiring page lock, then the page gets switched from
filecache to swapcache (and ->mapping set to NULL) whatever the refcount.
Fault it back in to get the page->mapping needed for key->shared.inode.

Reported-by: Sasha Levin <levinsasha928@gmail.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agonetfilter: ctnetlink: fix timeout calculation
Xi Wang [Fri, 30 Dec 2011 15:40:17 +0000 (10:40 -0500)]
netfilter: ctnetlink: fix timeout calculation

The sanity check (timeout < 0) never works; the dividend is unsigned
and so is the division, which should have been a signed division.

long timeout = (ct->timeout.expires - jiffies) / HZ;
if (timeout < 0)
timeout = 0;

This patch converts the time values to signed for the division.

Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
12 years agoipvs: try also real server with port 0 in backup server
Julian Anastasov [Fri, 30 Dec 2011 05:19:02 +0000 (14:19 +0900)]
ipvs: try also real server with port 0 in backup server

We should not forget to try for real server with port 0
in the backup server when processing the sync message. We should
do it in all cases because the backup server can use different
forwarding method.

Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
12 years agoskge: restore rx multicast filter on resume and after config changes
Florian Zumbiehl [Fri, 30 Dec 2011 17:30:09 +0000 (17:30 +0000)]
skge: restore rx multicast filter on resume and after config changes

Restore skge hardware registers for multicast filtering to their
appropriate values after system resume and after hardware restarts
that are done when changing certain settings.

Signed-off-by: Florian Zumbiehl <florz@florz.de>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: query link status in be_open()
Ajit Khaparde [Fri, 30 Dec 2011 12:15:40 +0000 (12:15 +0000)]
be2net: query link status in be_open()

be2net gets an async link status notification from the FW when it creates
an MCC queue. There are some cases in which this gratuitous notification
is not received from FW. To cover this explicitly query the link status
in be_open().

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: fix range check for set_qos for a VF
Ajit Khaparde [Fri, 30 Dec 2011 12:15:30 +0000 (12:15 +0000)]
be2net: fix range check for set_qos for a VF

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: fix be_vlan_add/rem_vid
Ajit Khaparde [Fri, 30 Dec 2011 12:15:12 +0000 (12:15 +0000)]
be2net: fix be_vlan_add/rem_vid

1) fix be_vlan_add/rem_vid to return proper status
2) perform appropriate housekeeping if firmware command succeeds.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomlx4_en: nullify cq->vector field when closing completion queue
Yevgeny Petrilin [Thu, 29 Dec 2011 05:49:58 +0000 (05:49 +0000)]
mlx4_en: nullify cq->vector field when closing completion queue

Caused loss of connectivity when changing ring size.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetem: fix classful handling
Eric Dumazet [Wed, 28 Dec 2011 23:12:02 +0000 (23:12 +0000)]
netem: fix classful handling

Commit 10f6dfcfde (Revert "sch_netem: Remove classful functionality")
reintroduced classful functionality to netem, but broke basic netem
behavior :

netem uses an t(ime)fifo queue, and store timestamps in skb->cb[]

If qdisc is changed, time constraints are not respected and other qdisc
can destroy skb->cb[] and block netem at dequeue time.

Fix this by always using internal tfifo, and optionally attach a child
qdisc to netem (or a tree of qdiscs)

Example of use :

DEV=eth3
tc qdisc del dev $DEV root
tc qdisc add dev $DEV root handle 30: est 1sec 8sec netem delay 20ms 10ms
tc qdisc add dev $DEV handle 40:0 parent 30:0 tbf \
burst 20480 limit 20480 mtu 1514 rate 32000bps

qdisc netem 30: root refcnt 18 limit 1000 delay 20.0ms  10.0ms
 Sent 190792 bytes 413 pkt (dropped 0, overlimits 0 requeues 0)
 rate 18416bit 3pps backlog 0b 0p requeues 0
qdisc tbf 40: parent 30: rate 256000bit burst 20Kb/8 mpu 0b lat 0us
 Sent 190792 bytes 413 pkt (dropped 6, overlimits 10 requeues 0)
 backlog 0b 5p requeues 0

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoIPv6: Avoid taking write lock for /proc/net/ipv6_route
Josh Hunt [Wed, 28 Dec 2011 13:23:07 +0000 (13:23 +0000)]
IPv6: Avoid taking write lock for /proc/net/ipv6_route

During some debugging I needed to look into how /proc/net/ipv6_route
operated and in my digging I found its calling fib6_clean_all() which uses
"write_lock_bh(&table->tb6_lock)" before doing the walk of the table. I
found this on 2.6.32, but reading the code I believe the same basic idea
exists currently. Looking at the rtnetlink code they are only calling
"read_lock_bh(&table->tb6_lock);" via fib6_dump_table(). While I realize
reading from proc isn't the recommended way of fetching the ipv6 route
table; taking a write lock seems unnecessary and would probably cause
network performance issues.

To verify this I loaded up the ipv6 route table and then ran iperf in 3
cases:
  * doing nothing
  * reading ipv6 route table via proc
    (while :; do cat /proc/net/ipv6_route > /dev/null; done)
  * reading ipv6 route table via rtnetlink
    (while :; do ip -6 route show table all > /dev/null; done)

* Load the ipv6 route table up with:
  * for ((i = 0;i < 4000;i++)); do ip route add unreachable 2000::$i; done

* iperf commands:
  * client: iperf -i 1 -V -c <ipv6 addr>
  * server: iperf -V -s

* iperf results - 3 runs each (in Mbits/sec)
  * nothing: client: 927,927,927 server: 927,927,927
  * proc: client: 179,97,96,113 server: 142,112,133
  * iproute: client: 928,927,928 server: 927,927,927

lock_stat shows taking the write lock is causing the slowdown. Using this
info I decided to write a version of fib6_clean_all() which replaces
write_lock_bh(&table->tb6_lock) with read_lock_bh(&table->tb6_lock). With
this new function I see the same results as with my rtnetlink iperf test.

Signed-off-by: Josh Hunt <joshhunt00@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agounix_diag: Fixup RQLEN extension report
Pavel Emelyanov [Fri, 30 Dec 2011 00:54:39 +0000 (00:54 +0000)]
unix_diag: Fixup RQLEN extension report

While it's not too late fix the recently added RQLEN diag extension
to report rqlen and wqlen in the same way as TCP does.

I.e. for listening sockets the ack backlog length (which is the input
queue length for socket) in rqlen and the max ack backlog length in
wqlen, and what the CINQ/OUTQ ioctls do for established.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoaf_unix: Move CINQ/COUTQ code to helpers
Pavel Emelyanov [Fri, 30 Dec 2011 00:54:11 +0000 (00:54 +0000)]
af_unix: Move CINQ/COUTQ code to helpers

Currently tcp diag reports rqlen and wqlen values similar to how
the CINQ/COUTQ iotcls do. To make unix diag report these values
in the same way move the respective code into helpers.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'fixes' of http://ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur...
Linus Torvalds [Fri, 30 Dec 2011 21:45:34 +0000 (13:45 -0800)]
Merge branch 'fixes' of ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur/linux-2.6-arm

* 'fixes' of http://ftp.arm.linux.org.uk/pub/linux/arm/kernel/git-cur/linux-2.6-arm:
  ARM: 7237/1: PL330: Fix driver freeze
  ARM: 7197/1: errata: Remove SMP dependency for erratum 751472
  ARM: 7196/1: errata: Remove SMP dependency for erratum 720789
  ARM: 7220/1: mmc: mmci: Fixup error handling for dma
  ARM: 7214/1: mmc: mmci: Fixup handling of MCI_STARTBITERR

12 years agounix_diag: Add the MEMINFO extension
Pavel Emelyanov [Fri, 30 Dec 2011 09:27:43 +0000 (09:27 +0000)]
unix_diag: Add the MEMINFO extension

[ Fix indentation of sock_diag*() calls. -DaveM ]

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc
Linus Torvalds [Fri, 30 Dec 2011 21:43:45 +0000 (13:43 -0800)]
Merge branch 'fixes' of git://git./linux/kernel/git/arm/arm-soc

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  ARM: plat-orion: make gpiochip label unique
  enable uncompress log on cpuimx35sd
  cpuimx35: fix touchscreen support
  cpuimx35sd: fix Kconfig
  clock-imx35: fix reboot in internal boot mode
  dma: MX3_IPU fix depends
  imx_v4_v5_defconfig: update default configuration
  cpuimx25sd: fix Kconfig
  arm/imx: fix cpufreq section mismatch
  ARM:imx:fix pwm period value
  ARM: OMAP: hwmod data: fix iva and mailbox hwmods for OMAP 3

12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Fri, 30 Dec 2011 21:42:41 +0000 (13:42 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: sentelic - fix retrieving number of buttons
  Input: sentelic - release mutex upon register write failure

12 years agoinet_diag: Add the SKMEMINFO extension
Pavel Emelyanov [Fri, 30 Dec 2011 00:53:32 +0000 (00:53 +0000)]
inet_diag: Add the SKMEMINFO extension

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosock_diag: Introduce the meminfo nla core (v2)
Pavel Emelyanov [Fri, 30 Dec 2011 00:53:13 +0000 (00:53 +0000)]
sock_diag: Introduce the meminfo nla core (v2)

Add a routine that dumps memory-related values of a socket.
It's made as an array to make it possible to add more stuff
here later without breaking compatibility.

Since v1: The SK_MEMINFO_ constants are in userspace
visible part of sock_diag.h, the rest is under __KERNEL__.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agounix_diag: Include unix_diag.h into header-y target
Pavel Emelyanov [Fri, 30 Dec 2011 00:52:51 +0000 (00:52 +0000)]
unix_diag: Include unix_diag.h into header-y target

The headers check complains it should include the linux/types.h
withing, thus add this one.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosock_diag: Arrange sock_diag.h such that it is exportable to userspace
Pavel Emelyanov [Fri, 30 Dec 2011 00:52:21 +0000 (00:52 +0000)]
sock_diag: Arrange sock_diag.h such that it is exportable to userspace

Properly toss existing components around the ifdef __KERNEL__
and include the header into the header-y target.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
Linus Torvalds [Fri, 30 Dec 2011 21:34:22 +0000 (13:34 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sage/ceph-client

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  ceph: disable use of dcache for readdir etc.

12 years agoMerge branch 'v3.2-samsung-fixes-4' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 30 Dec 2011 21:34:00 +0000 (13:34 -0800)]
Merge branch 'v3.2-samsung-fixes-4' of git://git./linux/kernel/git/kgene/linux-samsung

* 'v3.2-samsung-fixes-4' of git://git.kernel.org/pub/scm/linux/kernel/git/kgene/linux-samsung:
  ARM: EXYNOS: Remove duplicated SROMC static memory mapping
  ARM: SAMSUNG: Fix build error when selecting CPU_FREQ_S3C24XX_DEBUGFS on S3C2440

12 years agoRevert "clockevents: Set noop handler in clockevents_exchange_device()"
Linus Torvalds [Fri, 30 Dec 2011 21:24:40 +0000 (13:24 -0800)]
Revert "clockevents: Set noop handler in clockevents_exchange_device()"

This reverts commit de28f25e8244c7353abed8de0c7792f5f883588c.

It results in resume problems for various people. See for example

  http://thread.gmane.org/gmane.linux.kernel/1233033
  http://thread.gmane.org/gmane.linux.kernel/1233389
  http://thread.gmane.org/gmane.linux.kernel/1233159
  http://thread.gmane.org/gmane.linux.kernel/1227868/focus=1230877

and the fedora and ubuntu bug reports

  https://bugzilla.redhat.com/show_bug.cgi?id=767248
  https://bugs.launchpad.net/ubuntu/+source/linux/+bug/904569

which got bisected down to the stable version of this commit.

Reported-by: Jonathan Nieder <jrnieder@gmail.com>
Reported-by: Phil Miller <mille121@illinois.edu>
Reported-by: Philip Langdale <philipl@overt.org>
Reported-by: Tim Gardner <tim.gardner@canonical.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Greg KH <gregkh@suse.de>
Cc: stable@kernel.org # for stable kernels that applied the original
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge git://www.linux-watchdog.org/linux-watchdog
Linus Torvalds [Fri, 30 Dec 2011 20:13:03 +0000 (12:13 -0800)]
Merge git://www.linux-watchdog.org/linux-watchdog

* git://www.linux-watchdog.org/linux-watchdog:
  watchdog: iTCO_wdt.c - problems with newer hardware due to SMI clearing (part 2)
  watchdog: hpwdt: Changes to handle NX secure bit in 32bit path
  watchdog: sp805: Fix section mismatch in ID table.
  watchdog: move coh901327 state holders

12 years agoMerge branch 'tipc-Dec29-2011' of git://git.kernel.org/pub/scm/linux/kernel/git/paulg...
David S. Miller [Fri, 30 Dec 2011 18:46:54 +0000 (13:46 -0500)]
Merge branch 'tipc-Dec29-2011' of git://git./linux/kernel/git/paulg/linux

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Fri, 30 Dec 2011 18:04:14 +0000 (13:04 -0500)]
Merge git://git./linux/kernel/git/davem/net

12 years agotipc: rename struct bearer_name to struct tipc_bearer_names
Paul Gortmaker [Fri, 30 Dec 2011 02:39:49 +0000 (21:39 -0500)]
tipc: rename struct bearer_name to struct tipc_bearer_names

The addition of the "s" to indicate pluralization is intentional,
since the struct actually contains two name variants.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct link* to struct tipc_link*
Paul Gortmaker [Fri, 30 Dec 2011 01:58:42 +0000 (20:58 -0500)]
tipc: rename struct link* to struct tipc_link*

This converts the following:

struct link -> struct tipc_link
struct link_req -> struct tipc_link_req
struct link_name -> struct tipc_link_name

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct bcbearer* to tipc_bcbearer*
Paul Gortmaker [Fri, 30 Dec 2011 01:55:27 +0000 (20:55 -0500)]
tipc: rename struct bcbearer* to tipc_bcbearer*

This changes both the struct bcbearer and struct bcbearer_pair to
have the "tipc_" prefix.  Runtime behaviour is unchanged.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct bclink to struct tipc_bclink
Paul Gortmaker [Fri, 30 Dec 2011 01:52:18 +0000 (20:52 -0500)]
tipc: rename struct bclink to struct tipc_bclink

Make this rename so that it is consistent with the majority
of the other tipc structs and to assist in removing any
ambiguity with other similar names in other subsystems.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct subscriber to struct tipc_subscriber
Paul Gortmaker [Fri, 30 Dec 2011 01:49:39 +0000 (20:49 -0500)]
tipc: rename struct subscriber to struct tipc_subscriber

Make this rename so that it is consistent with the majority
of the other tipc structs and to assist in removing any
ambiguity with other similar names in other subsystems.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct subscription to struct tipc_subscription
Paul Gortmaker [Fri, 30 Dec 2011 01:43:44 +0000 (20:43 -0500)]
tipc: rename struct subscription to struct tipc_subscription

Make this rename so that it is consistent with the majority
of the other tipc structs and to assist in removing any
ambiguity with other similar names in other subsystems.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct port_list to struct tipc_port_list
Paul Gortmaker [Fri, 30 Dec 2011 01:33:30 +0000 (20:33 -0500)]
tipc: rename struct port_list to struct tipc_port_list

Make this rename so that it is consistent with the majority
of the other tipc structs and to assist in removing any
ambiguity with other similar names in other subsystems.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agotipc: rename struct media to struct tipc_media
Paul Gortmaker [Fri, 30 Dec 2011 01:19:42 +0000 (20:19 -0500)]
tipc: rename struct media to struct tipc_media

Give it a meaningful prefix, as suggested by DaveM, so that it
is consistent with things like struct tipc_bearer, and so it isn't
confused with anything else.  This has no impact on the actual
runtime code behaviour.

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
12 years agoMerge branch 'iommu/fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/joro...
Linus Torvalds [Fri, 30 Dec 2011 01:36:15 +0000 (17:36 -0800)]
Merge branch 'iommu/fixes' of git://git./linux/kernel/git/joro/iommu

* 'iommu/fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu: Initialize domain->handler in iommu_domain_alloc()

12 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Fri, 30 Dec 2011 01:35:33 +0000 (17:35 -0800)]
Merge git://git./linux/kernel/git/davem/net

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net:
  packet: fix possible dev refcnt leak when bind fail
  netem: dont call vfree() under spinlock and BH disabled
  netfilter: ctnetlink: fix scheduling while atomic if helper is autoloaded
  netfilter: ctnetlink: fix return value of ctnetlink_get_expect()

12 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 30 Dec 2011 01:09:16 +0000 (17:09 -0800)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86: Fix raw_spin_unlock_irqrestore() usage
  oprofile, arm/sh: Fix oprofile_arch_exit() linkage issue

12 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Fri, 30 Dec 2011 01:05:45 +0000 (17:05 -0800)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs:
  xfs: log all dirty inodes in xfs_fs_sync_fs
  xfs: log the inode in ->write_inode calls for kupdate

12 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-block
Linus Torvalds [Fri, 30 Dec 2011 00:33:37 +0000 (16:33 -0800)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block

* 'for-linus' of git://git.kernel.dk/linux-block:
  block: fix blk_queue_end_tag()
  block: re-use existing 'reading' variable instead of checking direction again
  block, cfq: fix empty queue crash caused by request merge

12 years agomm: hugetlb: fix non-atomic enqueue of huge page
Hillf Danton [Wed, 28 Dec 2011 23:57:16 +0000 (15:57 -0800)]
mm: hugetlb: fix non-atomic enqueue of huge page

If a huge page is enqueued under the protection of hugetlb_lock, then the
operation is atomic and safe.

Signed-off-by: Hillf Danton <dhillf@gmail.com>
Reviewed-by: Michal Hocko <mhocko@suse.cz>
Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: <stable@vger.kernel.org> [2.6.37+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoprocfs: do not confuse jiffies with cputime64_t
Andreas Schwab [Wed, 28 Dec 2011 23:57:15 +0000 (15:57 -0800)]
procfs: do not confuse jiffies with cputime64_t

Commit 2a95ea6c0d129b4 ("procfs: do not overflow get_{idle,iowait}_time
for nohz") did not take into account that one some architectures jiffies
and cputime use different units.

This causes get_idle_time() to return numbers in the wrong units, making
the idle time fields in /proc/stat wrong.

Instead of converting the usec value returned by
get_cpu_{idle,iowait}_time_us to units of jiffies, use the new function
usecs_to_cputime64 to convert it to the correct unit of cputime64_t.

Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Acked-by: Michal Hocko <mhocko@suse.cz>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: "Artem S. Tashkinov" <t.artem@mailcity.com>
Cc: Dave Jones <davej@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agomm/mempolicy.c: refix mbind_range() vma issue
KOSAKI Motohiro [Wed, 28 Dec 2011 23:57:11 +0000 (15:57 -0800)]
mm/mempolicy.c: refix mbind_range() vma issue

commit 8aacc9f550 ("mm/mempolicy.c: fix pgoff in mbind vma merge") is the
slightly incorrect fix.

Why? Think following case.

1. map 4 pages of a file at offset 0

   [0123]

2. map 2 pages just after the first mapping of the same file but with
   page offset 2

   [0123][23]

3. mbind() 2 pages from the first mapping at offset 2.
   mbind_range() should treat new vma is,

   [0123][23]
     |23|
     mbind vma

   but it does

   [0123][23]
     |01|
     mbind vma

   Oops. then, it makes wrong vma merge and splitting ([01][0123] or similar).

This patch fixes it.

[testcase]
  test result - before the patch

case4: 126: test failed. expect '2,4', actual '2,2,2'
        case5: passed
case6: passed
case7: passed
case8: passed
case_n: 246: test failed. expect '4,2', actual '1,4'

------------[ cut here ]------------
kernel BUG at mm/filemap.c:135!
invalid opcode: 0000 [#4] SMP DEBUG_PAGEALLOC

(snip long bug on messages)

  test result - after the patch

case4: passed
        case5: passed
case6: passed
case7: passed
case8: passed
case_n: passed

  source:  mbind_vma_test.c
============================================================
 #include <numaif.h>
 #include <numa.h>
 #include <sys/mman.h>
 #include <stdio.h>
 #include <unistd.h>
 #include <stdlib.h>
 #include <string.h>

static unsigned long pagesize;
void* mmap_addr;
struct bitmask *nmask;
char buf[1024];
FILE *file;
char retbuf[10240] = "";
int mapped_fd;

char *rubysrc = "ruby -e '\
  pid = %d; \
  vstart = 0x%llx; \
  vend = 0x%llx; \
  s = `pmap -q #{pid}`; \
  rary = []; \
  s.each_line {|line|; \
    ary=line.split(\" \"); \
    addr = ary[0].to_i(16); \
    if(vstart <= addr && addr < vend) then \
      rary.push(ary[1].to_i()/4); \
    end; \
  }; \
  print rary.join(\",\"); \
'";

void init(void)
{
void* addr;
char buf[128];

nmask = numa_allocate_nodemask();
numa_bitmask_setbit(nmask, 0);

pagesize = getpagesize();

sprintf(buf, "%s", "mbind_vma_XXXXXX");
mapped_fd = mkstemp(buf);
if (mapped_fd == -1)
perror("mkstemp "), exit(1);
unlink(buf);

if (lseek(mapped_fd, pagesize*8, SEEK_SET) < 0)
perror("lseek "), exit(1);
if (write(mapped_fd, "\0", 1) < 0)
perror("write "), exit(1);

addr = mmap(NULL, pagesize*8, PROT_NONE,
    MAP_SHARED, mapped_fd, 0);
if (addr == MAP_FAILED)
perror("mmap "), exit(1);

if (mprotect(addr+pagesize, pagesize*6, PROT_READ|PROT_WRITE) < 0)
perror("mprotect "), exit(1);

mmap_addr = addr + pagesize;

/* make page populate */
memset(mmap_addr, 0, pagesize*6);
}

void fin(void)
{
void* addr = mmap_addr - pagesize;
munmap(addr, pagesize*8);

memset(buf, 0, sizeof(buf));
memset(retbuf, 0, sizeof(retbuf));
}

void mem_bind(int index, int len)
{
int err;

err = mbind(mmap_addr+pagesize*index, pagesize*len,
    MPOL_BIND, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}

void mem_interleave(int index, int len)
{
int err;

err = mbind(mmap_addr+pagesize*index, pagesize*len,
    MPOL_INTERLEAVE, nmask->maskp, nmask->size, 0);
if (err)
perror("mbind "), exit(err);
}

void mem_unbind(int index, int len)
{
int err;

err = mbind(mmap_addr+pagesize*index, pagesize*len,
    MPOL_DEFAULT, NULL, 0, 0);
if (err)
perror("mbind "), exit(err);
}

void Assert(char *expected, char *value, char *name, int line)
{
if (strcmp(expected, value) == 0) {
fprintf(stderr, "%s: passed\n", name);
return;
}
else {
fprintf(stderr, "%s: %d: test failed. expect '%s', actual '%s'\n",
name, line,
expected, value);
// exit(1);
}
}

/*
      AAAA
    PPPPPPNNNNNN
    might become
    PPNNNNNNNNNN
    case 4 below
*/
void case4(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);

mem_bind(0, 4);
mem_unbind(2, 2);

file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case4", __LINE__);

fin();
}

/*
       AAAA
 PPPPPPNNNNNN
 might become
 PPPPPPPPPPNN
 case 5 below
*/
void case5(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);

mem_bind(0, 2);
mem_bind(2, 2);

file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case5", __LINE__);

fin();
}

/*
    AAAA
PPPPNNNNXXXX
might become
PPPPPPPPPPPP 6
*/
void case6(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);

mem_bind(0, 2);
mem_bind(4, 2);
mem_bind(2, 2);

file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("6", retbuf, "case6", __LINE__);

fin();
}

/*
    AAAA
PPPPNNNNXXXX
might become
PPPPPPPPXXXX 7
*/
void case7(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);

mem_bind(0, 2);
mem_interleave(4, 2);
mem_bind(2, 2);

file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case7", __LINE__);

fin();
}

/*
    AAAA
PPPPNNNNXXXX
might become
PPPPNNNNNNNN 8
*/
void case8(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);

mem_bind(0, 2);
mem_interleave(4, 2);
mem_interleave(2, 2);

file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("2,4", retbuf, "case8", __LINE__);

fin();
}

void case_n(void)
{
init();
sprintf(buf, rubysrc, getpid(), mmap_addr, mmap_addr+pagesize*6);

/* make redundunt mappings [0][1234][34][7] */
mmap(mmap_addr + pagesize*4, pagesize*2, PROT_READ|PROT_WRITE,
     MAP_FIXED|MAP_SHARED, mapped_fd, pagesize*3);

/* Expect to do nothing. */
mem_unbind(2, 2);

file = popen(buf, "r");
fread(retbuf, sizeof(retbuf), 1, file);
Assert("4,2", retbuf, "case_n", __LINE__);

fin();
}

int main(int argc, char** argv)
{
case4();
case5();
case6();
case7();
case8();
case_n();

return 0;
}
=============================================================

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Minchan Kim <minchan.kim@gmail.com>
Cc: Caspar Zhang <caspar@casparzhang.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Lee Schermerhorn <lee.schermerhorn@hp.com>
Cc: <stable@vger.kernel.org> [3.1.x]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agogspca: Fix bulk mode cameras no longer working (regression fix)
Hans de Goede [Thu, 29 Dec 2011 21:09:21 +0000 (19:09 -0200)]
gspca: Fix bulk mode cameras no longer working (regression fix)

The new iso bandwidth calculation code accidentally has broken support
for bulk mode cameras. This has broken the following drivers:
finepix, jeilinj, ovfx2, ov534, ov534_9, se401, sq905, sq905c, sq930x,
stv0680, vicam.

Thix patch fixes this. Fix tested with: se401, sq905, sq905c, stv0680 & vicam
cams.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoipv6: Fix neigh lookup using NULL device.
David S. Miller [Thu, 29 Dec 2011 23:51:57 +0000 (18:51 -0500)]
ipv6: Fix neigh lookup using NULL device.

In some of the rt6_bind_neighbour() call sites, it hasn't hooked
up the rt->dst.dev pointer yet, so we'd deref a NULL pointer when
obtaining dev->ifindex for the neighbour hash function computation.

Just pass the netdevice explicitly in to fix this problem.

Reported-by: Bjarke Istrup Pedersen <gurligebis@gentoo.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agovirtio_net: use non-reentrant workqueue.
Rusty Russell [Thu, 29 Dec 2011 00:43:15 +0000 (00:43 +0000)]
virtio_net: use non-reentrant workqueue.

Michael S. Tsirkin also noticed that we could run the refill work
multiple CPUs: if we kick off a refill on one CPU and then on another,
they would both manipulate the queue at the same time (they use
napi_disable to avoid racing against the receive handler itself).

Tejun points out that this is what the WQ_NON_REENTRANT flag is for,
and that there is a convenient system kthread we can use.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agovirtio_net: set/cancel work on ndo_open/ndo_stop
Rusty Russell [Thu, 29 Dec 2011 00:42:38 +0000 (00:42 +0000)]
virtio_net: set/cancel work on ndo_open/ndo_stop

Michael S. Tsirkin noticed that we could run the refill work after
ndo_close, which can re-enable napi - we don't disable it until
virtnet_remove.  This is clearly wrong, so move the workqueue control
to ndo_open and ndo_stop (aka. virtnet_open and virtnet_close).

One subtle point: virtnet_probe() could simply fail if it couldn't
allocate a receive buffer, but that's less polite in virtnet_open() so
we schedule a refill as we do in the normal receive path if we run out
of memory.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv6: Report TCP timetstamp info in cacheinfo just like ipv4 does.
David S. Miller [Thu, 29 Dec 2011 20:22:33 +0000 (15:22 -0500)]
ipv6: Report TCP timetstamp info in cacheinfo just like ipv4 does.

I missed this while adding ipv6 support to inet_peer.

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomlx4_core: limiting VF port options
Yevgeny Petrilin [Thu, 29 Dec 2011 07:42:39 +0000 (07:42 +0000)]
mlx4_core: limiting VF port options

At the moment VFs can only operate in Eth mode.
In addition we don't want the VF to attempt link sensing,
so we block this option as well.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agomlx4_core: using array index for sense_allowed
Yevgeny Petrilin [Thu, 29 Dec 2011 07:42:34 +0000 (07:42 +0000)]
mlx4_core: using array index for sense_allowed

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosch_tbf: report backlog information
Eric Dumazet [Wed, 28 Dec 2011 23:27:44 +0000 (23:27 +0000)]
sch_tbf: report backlog information

Provide child qdisc backlog (byte count) information so that "tc -s
qdisc" can report it to user.

qdisc netem 30: root refcnt 18 limit 1000 delay 20.0ms  10.0ms
 Sent 948517 bytes 898 pkt (dropped 0, overlimits 0 requeues 1)
 rate 175056bit 16pps backlog 114b 1p requeues 1
qdisc tbf 40: parent 30: rate 256000bit burst 20Kb/8 mpu 0b lat 0us
 Sent 948517 bytes 898 pkt (dropped 15, overlimits 611 requeues 0)
 backlog 18168b 12p requeues 0

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetfilter: Kconfig: fix unmet xt_nfacct dependencies
Pablo Neira Ayuso [Wed, 28 Dec 2011 15:03:30 +0000 (15:03 +0000)]
netfilter: Kconfig: fix unmet xt_nfacct dependencies

warning: (NETFILTER_XT_MATCH_NFACCT) selects NETFILTER_NETLINK_ACCT which has
unmet direct dependencies (NET && INET && NETFILTER && NETFILTER_ADVANCED)

and then

ERROR: "nfnetlink_subsys_unregister" [net/netfilter/nfnetlink_acct.ko] undefined!
ERROR: "nfnetlink_subsys_register" [net/netfilter/nfnetlink_acct.ko] undefined!

Reported-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Acked-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoInput: sentelic - fix retrieving number of buttons
Tai-hwa Liang [Thu, 29 Dec 2011 17:47:36 +0000 (09:47 -0800)]
Input: sentelic - fix retrieving number of buttons

Fixing wrong register offset which is used to retrieve the number of buttons
attached to the hardware.

Signed-off-by: Tai-hwa Liang <avatar@sentelic.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
12 years agoceph: disable use of dcache for readdir etc.
Sage Weil [Thu, 29 Dec 2011 16:05:14 +0000 (08:05 -0800)]
ceph: disable use of dcache for readdir etc.

Ceph attempts to use the dcache to satisfy negative lookups and readdir
when the entire directory contents are in cache.  Disable this behavior
until lingering bugs in this code are shaken out; we'll re-enable these
hooks once things are fully stable.

Signed-off-by: Sage Weil <sage@newdream.net>
12 years agoblock: fix blk_queue_end_tag()
Dan Williams [Thu, 29 Dec 2011 08:16:28 +0000 (09:16 +0100)]
block: fix blk_queue_end_tag()

Commit 5e081591 "block: warn if tag is greater than real_max_depth"
cleaned up blk_queue_end_tag() to warn when the tag is truly invalid
(greater than real_max_depth).  However, it changed behavior in the tag <
max_depth case to not end the request.  Leading to triggering of
BUG_ON(blk_queued_rq(rq)) in the request completion path:

  http://marc.info/?l=linux-kernel&m=132204370518629&w=2

In order to allow blk_queue_resize_tags() to shrink the tag space
blk_queue_end_tag() must always complete tags with a value less than
real_max_depth regardless of the current max_depth.  The comment about
"handling the shrink case" seems to be what prompted changes in this
space, so remove it and BUG on all invalid tags (made even simpler by
Matthew's suggestion to use an unsigned compare).

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Cc: Tao Ma <boyu.mt@taobao.com>
Cc: Matthew Wilcox <matthew@wil.cx>
Reported-by: Meelis Roos <mroos@ut.ee>
Reported-by: Ed Nadolski <edmund.nadolski@intel.com>
Cc: Tejun Heo <tj@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
12 years agoipv6: Kill rt6i_dev and rt6i_expires defines.
David S. Miller [Thu, 29 Dec 2011 01:19:20 +0000 (20:19 -0500)]
ipv6: Kill rt6i_dev and rt6i_expires defines.

It just obscures that the netdevice pointer and the expires value are
implemented in the dst_entry sub-object of the ipv6 route.

And it makes grepping for dst_entry member uses much harder too.

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv6: Create fast inline ipv6 neigh lookup just like ipv4.
David S. Miller [Wed, 28 Dec 2011 20:41:23 +0000 (15:41 -0500)]
ipv6: Create fast inline ipv6 neigh lookup just like ipv4.

Also, create and use an rt6_bind_neighbour() in net/ipv6/route.c to
consolidate some common logic.

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv6: Use universal hash for NDISC.
David S. Miller [Wed, 28 Dec 2011 20:06:58 +0000 (15:06 -0500)]
ipv6: Use universal hash for NDISC.

In order to perform a proper universal hash on a vector of integers,
we have to use different universal hashes on each vector element.

Which means we need 4 different hash randoms for ipv6.

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetrom: avoid overflows in nr_setsockopt()
Xi Wang [Tue, 27 Dec 2011 09:44:53 +0000 (09:44 +0000)]
netrom: avoid overflows in nr_setsockopt()

Check setsockopt arguments to avoid overflows and return -EINVAL for
too large arguments.

Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoax25: avoid overflows in ax25_setsockopt()
Xi Wang [Tue, 27 Dec 2011 09:43:19 +0000 (09:43 +0000)]
ax25: avoid overflows in ax25_setsockopt()

Commit be639ac6 ("NET: AX.25: Check ioctl arguments to avoid overflows
further down the road") rejects very large arguments, but doesn't
completely fix overflows on 64-bit systems.  Consider the AX25_T2 case.

int opt;
...
if (opt < 1 || opt > ULONG_MAX / HZ) {
res = -EINVAL;
break;
}
ax25->t2 = opt * HZ;

The 32-bit multiplication opt * HZ would overflow before being assigned
to 64-bit ax25->t2.  This patch changes "opt" to unsigned long.

Signed-off-by: Xi Wang <xi.wang@gmail.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agogenetlink: add auto module loading
Stephen Hemminger [Wed, 28 Dec 2011 18:48:55 +0000 (13:48 -0500)]
genetlink: add auto module loading

When testing L2TP support, I discovered that the l2tp module is not autoloaded
as are other netlink interfaces. There is because of lack of hook in genetlink to call
request_module and load the module.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv6: Remove optimistic DAD flag test in ipv6_add_addr()
David Miller [Tue, 27 Dec 2011 09:53:05 +0000 (09:53 +0000)]
ipv6: Remove optimistic DAD flag test in ipv6_add_addr()

The route we have here is for the address being added to the interface,
ie. for input packet processing.

Therefore using that route to determine whether an output nexthop gateway
is known and resolved doesn't make any sense.

So, simply remove this test, it never triggered anyways.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-By: Neil Horman <nhorman@tuxdriver.com>
12 years agoMerge branch 'nf-next' of git://1984.lsi.us.es/net-next
David S. Miller [Wed, 28 Dec 2011 18:16:59 +0000 (13:16 -0500)]
Merge branch 'nf-next' of git://1984.lsi.us.es/net-next

12 years agonet: fec: Adjust ENET MDIO timeouts
Rogerio Pimentel [Tue, 27 Dec 2011 19:07:37 +0000 (14:07 -0500)]
net: fec: Adjust ENET MDIO timeouts

On extensive NFS boots on a mx6qsabrelite board it was noted that "FEC: MDIO read timeout" were occuring,
which caused failure on loading the FEC driver.

The original FEC_MII_TIMEOUT was set to 1 ms, which is too low when passed to the usecs_to_jiffies macro.

On ARM one jiffy is 10ms, so use a timeout of 30ms, which corresponds to 3 jiffies.

After running extensive NFS boots, the MDIO timeouts do not occur anymore with this change.

Signed-off-by: Rogerio Pimentel <rogerio.pimentel@freescale.com>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoARM: EXYNOS: Remove duplicated SROMC static memory mapping
Thomas Abraham [Wed, 28 Dec 2011 06:07:32 +0000 (15:07 +0900)]
ARM: EXYNOS: Remove duplicated SROMC static memory mapping

SROMC static memory mapping is included in the common s5p initialization
code. Hence, remove the duplicated SROMC static memory mapping for EXYNOS.

Signed-off-by: Thomas Abraham <thomas.abraham@linaro.org>
Cc: stable@kernel.org
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
12 years agoARM: SAMSUNG: Fix build error when selecting CPU_FREQ_S3C24XX_DEBUGFS on S3C2440
Denis Kuzmenko [Wed, 28 Dec 2011 05:04:51 +0000 (14:04 +0900)]
ARM: SAMSUNG: Fix build error when selecting CPU_FREQ_S3C24XX_DEBUGFS on S3C2440

Following is happened when CONFIG_CPU_FREQ_S3C24XX_DEBUGFS
is selected without building of s3c2410-iotiming.c file:

arch/arm/mach-s3c2440/built-in.o:(.data+0x38c): undefined reference to `s3c2410_iotiming_debugfs

Basically, the CONFIG_S3C2410_IOTIMING is not selected for
MACH_MINI2440. Because the s3c2410-iotiming.c is not ever
compiled and enabling CONFIG_CPU_FREQ_S3C24XX_DEBUGFS option
caused undefined reference to s3c2410_iotiming_debugfs()
defined in that file.  The s3c2410_iotiming_debugfs defined
as NULL for this case.

Signed-off-by: Denis Kuzmenko <linux@solonet.org.ua>
Cc: stable@kernel.org
[kgene.kim@samsung.com: removed useless changes]
Signed-off-by: Kukjin Kim <kgene.kim@samsung.com>
12 years agopacket: fix possible dev refcnt leak when bind fail
Wei Yongjun [Wed, 28 Dec 2011 03:32:41 +0000 (22:32 -0500)]
packet: fix possible dev refcnt leak when bind fail

If bind is fail when bind is called after set PACKET_FANOUT
sock option, the dev refcnt will leak.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetfilter: provide config option to disable ancient procfs parts
Jan Engelhardt [Thu, 21 Apr 2011 07:32:45 +0000 (09:32 +0200)]
netfilter: provide config option to disable ancient procfs parts

Using /proc/net/nf_conntrack has been deprecated in favour of the
conntrack(8) tool.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>