GitHub/exynos8895/android_kernel_samsung_universal8895.git
15 years agoiwlwifi: check return value of pci_enable_device
Winkler, Tomas [Tue, 27 Jan 2009 22:27:55 +0000 (14:27 -0800)]
iwlwifi: check return value of pci_enable_device

pci_enable_device is tagged with __must_check therefore
don't ignore the return value in pci_resume handlers

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoiwlwifi: iwl3945_send_tx_power must be static
Winkler, Tomas [Tue, 27 Jan 2009 22:27:54 +0000 (14:27 -0800)]
iwlwifi: iwl3945_send_tx_power must be static

iwl3945_send_tx_power must be static

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoiwlwifi: suppress unused variable warning when compiling w/o IWLWIFI_DEBUG
Winkler, Tomas [Tue, 27 Jan 2009 22:27:53 +0000 (14:27 -0800)]
iwlwifi: suppress unused variable warning when compiling w/o IWLWIFI_DEBUG

This patch adds __maybe_unused attribute to priv variables used in
functions that used it solely for debug printouts

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Cc: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoiwl3945: Getting rid of the *39_rxon iwl_priv fields
Samuel Ortiz [Tue, 27 Jan 2009 22:27:52 +0000 (14:27 -0800)]
iwl3945: Getting rid of the *39_rxon iwl_priv fields

The iwl_rxon_cmd is really just a iwl3945_rxon_cmd structure extension.
So, we can use the *_rxon fields from iwl_priv instead of the 3945 specific
ones (*39_rxon). We have to then be careful when submitting REPLY_RXON host
commands, since the command length as to be set according to the HW. As
another precaution the reserved4 and reserved5 fields are cleared before being
sent to the 3945.

With the *39_rxon removal, a lot of duplicated code can be removed from the
3945 code base.

Signed-off-by: Samuel Ortiz <samuel.ortiz@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoorinoco: checkpatch cleanup
Andrey Borzenkov [Sun, 25 Jan 2009 20:08:43 +0000 (23:08 +0300)]
orinoco: checkpatch cleanup

Fix errors and obvious warnings reported by checkpatch in all files
except orinoco.c. Orinoco.c is part of different patch series of Dave.

Signed-off-by: Andrey Borzenkov <arvidjaar@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agogro: Optimise TCP packet reception
Herbert Xu [Sun, 8 Feb 2009 18:00:40 +0000 (18:00 +0000)]
gro: Optimise TCP packet reception

gro: Optimise TCP packet reception

As this function can be called more than half a million times for
10GbE, it's important to optimise it as much as we can.

This patch uses bit ops to logical ops, as well as open coding
memcmp to exploit alignment properties.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogro: Optimise IPv4 packet reception
Herbert Xu [Sun, 8 Feb 2009 18:00:39 +0000 (18:00 +0000)]
gro: Optimise IPv4 packet reception

As this function can be called more than half a million times for
10GbE, it's important to optimise it as much as we can.

This patch does some obvious changes to use 2-byte and 4-byte
operations instead of byte-oriented ones where possible.  Bit
ops are also used to replace logical ops to reduce branching.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogro: Optimise Ethernet header comparison
Herbert Xu [Sun, 8 Feb 2009 18:00:37 +0000 (18:00 +0000)]
gro: Optimise Ethernet header comparison

This patch optimises the Ethernet header comparison to use 2-byte
and 4-byte xors instead of memcmp.  In order to facilitate this,
the actual comparison is now carried out by the callers of the
shared dev_gro_receive function.

This has a significant impact when receiving 1500B packets through
10GbE.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogro: Remember number of held packets instead of counting every time
Herbert Xu [Sun, 8 Feb 2009 18:00:36 +0000 (18:00 +0000)]
gro: Remember number of held packets instead of counting every time

This patch prepares for the move of the same_flow checks out of
dev_gro_receive.  As such we need to remember the number of held
packets since doing a loop just to count them every time is silly.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Increase default NET_SKB_PAD to 32.
David S. Miller [Mon, 9 Feb 2009 03:24:13 +0000 (19:24 -0800)]
net: Increase default NET_SKB_PAD to 32.

Several devices need to insert some "pre headers" in front of the
main packet data when they transmit a packet.

Currently we allocate only 16 bytes of pad room and this ends up not
being enough for some types of hardware (NIU, usb-net, s390 qeth,
etc.)

So increase this to 32.

Note that drivers still need to check in their transmit routine
whether enough headroom exists, and if not use skb_realloc_headroom().
Tunneling, IPSEC, and other encapsulation methods can cause the
padding area to be used up.

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Sat, 7 Feb 2009 10:52:44 +0000 (02:52 -0800)]
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

Conflicts:
drivers/net/wireless/iwlwifi/iwl-agn.c
drivers/net/wireless/iwlwifi/iwl3945-base.c

15 years agoigb: remove dead code in transmit routine
Jesse Brandeburg [Fri, 6 Feb 2009 23:23:32 +0000 (23:23 +0000)]
igb: remove dead code in transmit routine

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: update version number and copyright dates
Alexander Duyck [Fri, 6 Feb 2009 23:23:12 +0000 (23:23 +0000)]
igb: update version number and copyright dates

Update the version number to 1.3.16 and update copyright dates for 2009.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: fix two minor items found during code review
Alexander Duyck [Fri, 6 Feb 2009 23:22:52 +0000 (23:22 +0000)]
igb: fix two minor items found during code review

This patch addresses two minor items I found while cleaning up the igb
driver for our sourceforge version.

The first clears the context index if we don't flag that we need it.

The second item is that eims_other should be used instead of bit defines
when setting all of the EICS bits prior to reset.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: update stats before doing reset in igb_down
Alexander Duyck [Fri, 6 Feb 2009 23:22:32 +0000 (23:22 +0000)]
igb: update stats before doing reset in igb_down

It was seen with repeated interface up/down testing that there was a large
stray between the stats reported by the queues and the stats reported by the
HW.  It was found to be an issue in that hw stats were being reset without
first being recorded.  This change records the stats before wiping them from
the system via the reset.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: remove redundant count set and err_hw_init
Alexander Duyck [Fri, 6 Feb 2009 23:22:11 +0000 (23:22 +0000)]
igb: remove redundant count set and err_hw_init

Remove the setting of ring->count variables from igb_probe as they are
duplicating the same configuration that is done igb_alloc_queues.

Remove the err_hw_init tag as it can be replaced by err_sw_init.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: remove disable_av variable from mac_info struct
Alexander Duyck [Fri, 6 Feb 2009 23:21:51 +0000 (23:21 +0000)]
igb: remove disable_av variable from mac_info struct

The disable_av variable is never used by the driver and provides no value as
it is likely a leftover debugging variable.  I have removed it and replaced
the one spot that checked for it with a check for a valid address.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: change pba size determination from if to switch statement
Alexander Duyck [Fri, 6 Feb 2009 23:21:31 +0000 (23:21 +0000)]
igb: change pba size determination from if to switch statement

As additional hardware is added to the igb driver it is easier to support
the expansion via switch statements instead of using nested ifs.  For
this reason I am changing this to a switch statement.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: move get_hw_control within igb_resume.
Alexander Duyck [Fri, 6 Feb 2009 23:21:10 +0000 (23:21 +0000)]
igb: move get_hw_control within igb_resume.

Move igb_get_hw_control up so that it is called just after the reset in
igb_resume.  This notifies the HW sooner that the driver is reassuming
control of the device.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: don't read eicr when responding to legacy interrupts
Alexander Duyck [Fri, 6 Feb 2009 23:20:49 +0000 (23:20 +0000)]
igb: don't read eicr when responding to legacy interrupts

The interrupt handler was reading eicr and then doing nothing with the
result.  I have removed the variable and the register read since they
provide no value to the legacy interrupt handler.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: remove unnecessary adapter->hw calls when just hw-> will do.
Alexander Duyck [Fri, 6 Feb 2009 23:20:31 +0000 (23:20 +0000)]
igb: remove unnecessary adapter->hw calls when just hw-> will do.

There were several spots in the code making calls to adapter->hw when they
could have just been accessing hw-> directly.  I cleaned up the spots where
this was visibly apparent.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: rename igb_update_mc_addr_list_82575 to not include the 82575
Alexander Duyck [Fri, 6 Feb 2009 23:20:10 +0000 (23:20 +0000)]
igb: rename igb_update_mc_addr_list_82575 to not include the 82575

There isn't much point in having the _82575 hanging off the end of this
function since there aren't any other version of this function running
around within this driver.  This also allows for a bit of whitespace
cleanup due to a shorter function name.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: remove redundant timer updates and cleanup watchdog_task
Alexander Duyck [Fri, 6 Feb 2009 23:19:50 +0000 (23:19 +0000)]
igb: remove redundant timer updates and cleanup watchdog_task

The igb watchdog task is modifying the watchdog timer twice duing a single
run.  It only needs to be called once to reschedule itself for 2 seconds from
the last time it ran.

In addition I removed the allocation of the mac_info structure since it is
only called twice and is easier to access via the e1000_hw struct.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: cleanup igb_netpoll to be more friendly with napi & GRO
Alexander Duyck [Fri, 6 Feb 2009 23:19:29 +0000 (23:19 +0000)]
igb: cleanup igb_netpoll to be more friendly with napi & GRO

This patch cleans up igb_netpoll so that it is more friendly with both the
current napi and newly introduced GRO features.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: add counter for dma out of sync errors
Alexander Duyck [Fri, 6 Feb 2009 23:19:08 +0000 (23:19 +0000)]
igb: add counter for dma out of sync errors

Add a counter for dma out of sync errors reported via interrupt.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: update testing done by ethtool
Alexander Duyck [Fri, 6 Feb 2009 23:18:48 +0000 (23:18 +0000)]
igb: update testing done by ethtool

Most of the code for the testing has pretty much become stale at this point
and is need of update.  This update just streamlines most of the code,
widens the range of interrupt testing.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: update feature flags supported in ethtool
Alexander Duyck [Fri, 6 Feb 2009 23:18:27 +0000 (23:18 +0000)]
igb: update feature flags supported in ethtool

This driver is currently using HW_CSUM which is not correct.  Update this
to use the IP_CSUM and IPV6_CSUM flags.  In addition consolidate the TSO
flag setting.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: remove unused rx_hdr_split statistic
Alexander Duyck [Fri, 6 Feb 2009 23:18:06 +0000 (23:18 +0000)]
igb: remove unused rx_hdr_split statistic

This statistic is not used and so it is safe to remove

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: rename nvm ops
Alexander Duyck [Fri, 6 Feb 2009 23:17:47 +0000 (23:17 +0000)]
igb: rename nvm ops

All of the nvm ops have the tag _nvm added to the end which is redundant
since all of the calls to the ops have to go through the nvm ops struct
anyway.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: rename phy ops
Alexander Duyck [Fri, 6 Feb 2009 23:17:26 +0000 (23:17 +0000)]
igb: rename phy ops

This patch renames write_phy_reg to write_reg and read_phy_reg to read_reg.
It seems redundant to call out phy in an operation that is part of the
phy_ops struct.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: read address from RAH/RAL instead of from EEPROM
Alexander Duyck [Fri, 6 Feb 2009 23:17:06 +0000 (23:17 +0000)]
igb: read address from RAH/RAL instead of from EEPROM

Instead of pulling the mac address from EEPROM it is easier to pull it from
the RAL/RAH registers and then just copy it into the address structures.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: make dev_spec a union and remove dynamic allocation
Alexander Duyck [Fri, 6 Feb 2009 23:16:45 +0000 (23:16 +0000)]
igb: make dev_spec a union and remove dynamic allocation

This patch makes dev_spec a union and simplifies it so that it does not
require dynamic allocation and freeing in the driver.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: add link check function
Alexander Duyck [Fri, 6 Feb 2009 23:16:24 +0000 (23:16 +0000)]
igb: add link check function

Add a link check function to contain all activities related to verifying
that the link is present.  The current approach is a bit cludgy and needs
to be cleaned up.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: remove check for needing an io port
Alexander Duyck [Fri, 6 Feb 2009 23:16:04 +0000 (23:16 +0000)]
igb: remove check for needing an io port

Since igb supports only pci-e nics and there is no plan to support any
legacy pci parts in the driver there isn't really much need for checking to
see if an io port is needed.

In the unlikely event that we do begin supporting legacy pci parts then we
can see about adding this code back to the driver.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: move initialization of number of queues into set_interrupt_capability
Alexander Duyck [Fri, 6 Feb 2009 23:15:45 +0000 (23:15 +0000)]
igb: move initialization of number of queues into set_interrupt_capability

This patch moves the initialization of the number of queues into
set_interrupt_capability.  This allows the number of queues to increase in
the unlikely event that the system initially fails to allocate enough msi-x
interrupts, does a suspend/resume, and then can allocate enough interrupts
on resume.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: move setting of buffsz out of repeated path in alloc_rx_buffers
Alexander Duyck [Fri, 6 Feb 2009 23:15:25 +0000 (23:15 +0000)]
igb: move setting of buffsz out of repeated path in alloc_rx_buffers

buffsz is being repeatedly set when allocaing buffers.  Since this value
should only need to be set once in the function I am moving it out of the
looped portion of the path.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: optimize/refactor receive path
Alexander Duyck [Fri, 6 Feb 2009 23:15:04 +0000 (23:15 +0000)]
igb: optimize/refactor receive path

While cleaning up the skb_over panic with small frames I found there was
room for improvement in the ordering of operations within the rx receive
flow.  These changes will place the prefetch for the next descriptor to a
point earlier in the rx path.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agosunhme: Don't match PCI devices in SBUS probe.
David S. Miller [Sat, 7 Feb 2009 10:20:25 +0000 (02:20 -0800)]
sunhme: Don't match PCI devices in SBUS probe.

Unfortunately, the OF device tree nodes for SBUS and PCI
hme devices have the same device node name on some systems.

So if the name of the parent node isn't 'sbus', skip it.

Based upon an excellent report and detective work by
Meelis Roos and Eric Brower.

Signed-off-by: David S. Miller <davem@davemloft.net>
Tested-by: Meelis Roos <mroos@linux.ee>
15 years agoixgbe: Fix a set_num_queues() bug that can result in num_(r|t)x_queues = 0
Peter P Waskiewicz Jr [Sat, 7 Feb 2009 10:16:59 +0000 (02:16 -0800)]
ixgbe: Fix a set_num_queues() bug that can result in num_(r|t)x_queues = 0

Now that our set_num_queues() routines for each feature are re-entrant, and
can be called at any point, they shouldn't zero out the feature's indices
or mask bits.  Subsequent calls into those routines for those features can
result in zero Rx and Tx queues being assigned, causing a panic later in
driver reinitialization.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: bump version to 63
Ayaz Abdulla [Sat, 7 Feb 2009 08:25:18 +0000 (00:25 -0800)]
forcedeth: bump version to 63

This patch bumps the version up to 63

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: recover error support
Ayaz Abdulla [Sat, 7 Feb 2009 08:25:00 +0000 (00:25 -0800)]
forcedeth: recover error support

This patch adds another type of recoverable error to the driver. It also
modifies the sequence for recovery to include a mac reset and clearing
of interrupts.

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: ethtool tx csum fix
Ayaz Abdulla [Sat, 7 Feb 2009 08:24:39 +0000 (00:24 -0800)]
forcedeth: ethtool tx csum fix

This patch fixes the ethtool tx csum "set" command. A recent patch was
submitted to remove HW_CSUM and use IP_CSUM instead. Therefore, the
corresponding ethtool command should also be modified.

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: msi interrupt fix
Ayaz Abdulla [Sat, 7 Feb 2009 08:24:15 +0000 (00:24 -0800)]
forcedeth: msi interrupt fix

This patch fixes an issue with the suspend/resume cycle with msi
interrupts. See bugzilla number 10487 for more details. The fix is to
re-setup a private msi pci config offset field.

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: mgmt unit interface
Ayaz Abdulla [Sat, 7 Feb 2009 08:23:57 +0000 (00:23 -0800)]
forcedeth: mgmt unit interface

This patch updates the logic used to communicate with the mgmt unit. It
also adds a version check for a newer mgmt unit firmware.

* Fixed udelay to schedule_timeout_uninterruptible

Signed-off-by: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet/sunrpc/xprtsock.c: some common code found
Ilpo Järvinen [Sat, 7 Feb 2009 07:48:33 +0000 (23:48 -0800)]
net/sunrpc/xprtsock.c: some common code found

$ diff-funcs xs_udp_write_space net/sunrpc/xprtsock.c
net/sunrpc/xprtsock.c xs_tcp_write_space
 --- net/sunrpc/xprtsock.c:xs_udp_write_space()
 +++ net/sunrpc/xprtsock.c:xs_tcp_write_space()
@@ -1,4 +1,4 @@
- * xs_udp_write_space - callback invoked when socket buffer space
+ * xs_tcp_write_space - callback invoked when socket buffer space
  *                             becomes available
  * @sk: socket whose state has changed
  *
@@ -7,12 +7,12 @@
  * progress, otherwise we'll waste resources thrashing kernel_sendmsg
  * with a bunch of small requests.
  */
-static void xs_udp_write_space(struct sock *sk)
+static void xs_tcp_write_space(struct sock *sk)
 {
  read_lock(&sk->sk_callback_lock);

- /* from net/core/sock.c:sock_def_write_space */
- if (sock_writeable(sk)) {
+ /* from net/core/stream.c:sk_stream_write_space */
+ if (sk_stream_wspace(sk) >= sk_stream_min_wspace(sk)) {
  struct socket *sock;
  struct rpc_xprt *xprt;

$ codiff net/sunrpc/xprtsock.o net/sunrpc/xprtsock.o.new
net/sunrpc/xprtsock.c:
  xs_tcp_write_space | -163
  xs_udp_write_space | -163
 2 functions changed, 326 bytes removed

net/sunrpc/xprtsock.c:
  xs_write_space | +179
 1 function changed, 179 bytes added

net/sunrpc/xprtsock.o.new:
 3 functions changed, 179 bytes added, 326 bytes removed, diff: -147

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6/addrconf: common code located
Ilpo Järvinen [Sat, 7 Feb 2009 07:48:01 +0000 (23:48 -0800)]
ipv6/addrconf: common code located

$ codiff net/ipv6/addrconf.o net/ipv6/addrconf.o.new
net/ipv6/addrconf.c:
 addrconf_notify | -267
1 function changed, 267 bytes removed

net/ipv6/addrconf.c:
 add_addr |  +86
1 function changed, 86 bytes added

net/ipv6/addrconf.o.new:
2 functions changed, 86 bytes added, 267 bytes removed, diff: -181

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6/ndisc: join error paths
Ilpo Järvinen [Sat, 7 Feb 2009 07:47:37 +0000 (23:47 -0800)]
ipv6/ndisc: join error paths

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoax25: more common return path joining
Ilpo Järvinen [Sat, 7 Feb 2009 07:47:14 +0000 (23:47 -0800)]
ax25: more common return path joining

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipmr: use goto to common label instead of opencoding
Ilpo Järvinen [Sat, 7 Feb 2009 07:46:51 +0000 (23:46 -0800)]
ipmr: use goto to common label instead of opencoding

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years ago9p: fix endian issues [attempt 3]
Eric Van Hensbergen [Sat, 7 Feb 2009 06:07:41 +0000 (22:07 -0800)]
9p: fix endian issues [attempt 3]

When the changes were done to the protocol last release, some endian
bugs crept in.  This patch fixes those endian problems and has been
verified to run on 32/64 bit and x86/ppc architectures.

This version of the patch incorporates the correct annotations
for endian variables.

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet_dma: call dmaengine_get only if NET_DMA enabled
David S. Miller [Sat, 7 Feb 2009 06:06:43 +0000 (22:06 -0800)]
net_dma: call dmaengine_get only if NET_DMA enabled

Based upon a patch from Atsushi Nemoto <anemo@mba.ocn.ne.jp>

--------------------
The commit 649274d993212e7c23c0cb734572c2311c200872 ("net_dma:
acquire/release dma channels on ifup/ifdown") added unconditional call
of dmaengine_get() to net_dma.  The API should be called only if
NET_DMA was enabled.
--------------------

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Dan Williams <dan.j.williams@intel.com>
15 years ago3c509: Fix resume from hibernation for PnP mode.
Ondrej Zary [Sat, 7 Feb 2009 06:04:08 +0000 (22:04 -0800)]
3c509: Fix resume from hibernation for PnP mode.

From: Ondrej Zary <linux@rainbow-software.org>

last year, I posted a patch which fixed hibernation on 3c509
cards. That was back in 2.6.24. It worked fine in 2.6.25. But then I
stopped using hibernation (as it did not work with my new IT8212 RAID
controller).

Now I fixed it and noticed that 3c509 does not wake up properly
anymore (in 2.6.28) - neither in PnP nor in ISA modes. ifconfig
down/up makes the card work again in PnP mode. However, in ISA mode,
ifconfig up ends with "No such device" error.

Comparing the 3c509 driver between 2.6.25 and 2.6.28, there's only
some statistics-related change. So the cause of the problem must be
somewhere else.

This patch makes the resume work in PnP mode, but it's still not
enough for ISA mode.

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agosungem: Soft lockup in sungem on Netra AC200 when switching interface up
Ilkka Virta [Sat, 7 Feb 2009 06:00:36 +0000 (22:00 -0800)]
sungem: Soft lockup in sungem on Netra AC200 when switching interface up

From: Ilkka Virta <itvirta@iki.fi>

In the lockup situation the driver seems to go off in an eternal storm
of interrupts right after calling request_irq(). It doesn't actually
do anything interesting in the interrupt handler. Since connecting the link
afterwards works, something later in initialization must fix this.

Looking at gem_do_start() and gem_open(), it seems that the only thing
done while opening the device after the request_irq(), is a call to
napi_enable().

I don't know what the ordering requirements are for the
initialization, but I boldly tried to move the napi_enable() call
inside gem_do_start() before the link state is checked and interrupts
subsequently enabled, and it seems to work for me. Doesn't even break
anything too obvious...

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoRxRPC: Fix a potential NULL dereference
David Howells [Sat, 7 Feb 2009 05:50:52 +0000 (21:50 -0800)]
RxRPC: Fix a potential NULL dereference

Fix a potential NULL dereference bug during error handling in
rxrpc_kernel_begin_call(), whereby rxrpc_put_transport() may be handed a NULL
pointer.

This was found with a code checker (http://repo.or.cz/w/smatch.git/).

Reported-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agor8169: Don't update statistics counters when interface is down
Ivan Vecera [Sat, 7 Feb 2009 05:49:57 +0000 (21:49 -0800)]
r8169: Don't update statistics counters when interface is down

Some Realtek chips (RTL8169sb/8110sb in my case) are unable to retrieve
ethtool statistics when the interface is down. The process stays in
endless loop in rtl8169_get_ethtool_stats. This is because these chips
need to have receiver enabled (CmdRxEnb bit in ChipCmd register) that is
cleared when the interface is going down. It's better to update statistics
only when the interface is up and otherwise return copy of statistics
grabbed when the interface was up (in rtl8169_close).

It is interesting that PCI-E NICs (like 8168b/8111b...) are not affected.

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoixgbe: Defeature Tx Head writeback
Peter P Waskiewicz Jr [Sat, 7 Feb 2009 05:47:24 +0000 (21:47 -0800)]
ixgbe: Defeature Tx Head writeback

Tx Head writeback is causing multi-microsecond stalls on PCIe chipsets, due
to partial cacheline writebacks.  Removing this feature removes these
issues.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoixgbe: Update flow control state machine in link setup
Peter P Waskiewicz Jr [Sat, 7 Feb 2009 05:46:54 +0000 (21:46 -0800)]
ixgbe: Update flow control state machine in link setup

The flow control handling is overly complicated and difficult to maintain.
This patch cleans up the flow control handling and makes it much more
explicit.  It also adds 1G flow control autonegotiation, for 1G copper
links, 1G KX links, and 1G fiber links.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoudp: Fix potential wrong ip_hdr(skb) pointers
Jesper Dangaard Brouer [Fri, 6 Feb 2009 09:59:12 +0000 (01:59 -0800)]
udp: Fix potential wrong ip_hdr(skb) pointers

Like the UDP header fix, pskb_may_pull() can potentially
alter the SKB buffer.  Thus the saddr and daddr, pointers
may point to the old skb->data buffer.

I haven't seen corruptions, as its only seen if the old
skb->data buffer were reallocated by another user and
written into very quickly (or poison'd by SLAB debugging).

Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: enable msix to default
Yinghai Lu [Fri, 6 Feb 2009 09:31:12 +0000 (01:31 -0800)]
forcedeth: enable msix to default

Impact: change default

msix and napic can work again

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: ck804 and mcp55 doesn't need timerirq
Yinghai Lu [Fri, 6 Feb 2009 09:30:56 +0000 (01:30 -0800)]
forcedeth: ck804 and mcp55 doesn't need timerirq

Impact: cleanup

so get less irq.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: disable irq at first before schedule rx
Yinghai Lu [Fri, 6 Feb 2009 09:30:36 +0000 (01:30 -0800)]
forcedeth: disable irq at first before schedule rx

Impact: clean up

schedule it later after disable it.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: don't clear nic_poll_irq too early
Yinghai Lu [Fri, 6 Feb 2009 09:30:01 +0000 (01:30 -0800)]
forcedeth: don't clear nic_poll_irq too early

Impact: fix bug

for msix, we still need that flag to enable irq respectively

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoforcedeth: make msi-x different name for rx-tx
Yinghai Lu [Fri, 6 Feb 2009 09:29:23 +0000 (01:29 -0800)]
forcedeth: make msi-x different name for rx-tx

Impact: make /proc/interrupts could show more info which irq is rx or other for msi-x

add three name fields for rx, tx, other

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoneigh: some entries can be skipped during dumping
Gautam Kachroo [Fri, 6 Feb 2009 08:52:04 +0000 (00:52 -0800)]
neigh: some entries can be skipped during dumping

neightbl_dump_info and neigh_dump_table  can skip entries if the
*fill*info functions return an error. This results in an incomplete
dump ((invoked by netlink requests for RTM_GETNEIGHTBL or
RTM_GETNEIGH)

nidx and idx should not be incremented if the current entry was not
placed in the output buffer

Signed-off-by: Gautam Kachroo <gk@aristanetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6: Disallow rediculious flowlabel option sizes.
David S. Miller [Fri, 6 Feb 2009 08:49:55 +0000 (00:49 -0800)]
ipv6: Disallow rediculious flowlabel option sizes.

Just like PKTINFO, limit the options area to 64K.

Based upon report by Eric Sesterhenn and analysis by
Roland Dreier.

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonetlink: change return-value logic of netlink_broadcast()
Pablo Neira Ayuso [Fri, 6 Feb 2009 07:56:36 +0000 (23:56 -0800)]
netlink: change return-value logic of netlink_broadcast()

Currently, netlink_broadcast() reports errors to the caller if no
messages at all were delivered:

1) If, at least, one message has been delivered correctly, returns 0.
2) Otherwise, if no messages at all were delivered due to skb_clone()
   failure, return -ENOBUFS.
3) Otherwise, if there are no listeners, return -ESRCH.

With this patch, the caller knows if the delivery of any of the
messages to the listeners have failed:

1) If it fails to deliver any message (for whatever reason), return
   -ENOBUFS.
2) Otherwise, if all messages were delivered OK, returns 0.
3) Otherwise, if no listeners, return -ESRCH.

In the current ctnetlink code and in Netfilter in general, we can add
reliable logging and connection tracking event delivery by dropping the
packets whose events were not successfully delivered over Netlink. Of
course, this option would be settable via /proc as this approach reduces
performance (in terms of filtered connections per seconds by a stateful
firewall) but providing reliable logging and event delivery (for
conntrackd) in return.

This patch also changes some clients of netlink_broadcast() that
may report ENOBUFS errors via printk. This error handling is not
of any help. Instead, the userspace daemons that are listening to
those netlink messages should resync themselves with the kernel-side
if they hit ENOBUFS.

BTW, netlink_broadcast() clients include those that call
cn_netlink_send(), nlmsg_multicast() and genlmsg_multicast() since they
internally call netlink_broadcast() and return its error value.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: normalize usage of serdes_has_link
Alex Chiang [Fri, 6 Feb 2009 07:55:45 +0000 (23:55 -0800)]
e1000e: normalize usage of serdes_has_link

Cosmetic change to use struct e1000_mac_info.serdes_has_link
consistently as the 'bool' that it's declared as.

No functional change.

Signed-off-by: Alex Chiang <achiang@hp.com>
Acked-by: Jeff Kirsher <Jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoixgbe: Display EEPROM version in ethtool -i queries
Peter P Waskiewicz Jr [Fri, 6 Feb 2009 07:54:42 +0000 (23:54 -0800)]
ixgbe: Display EEPROM version in ethtool -i queries

Currently ixgbe does not display the EEPROM version in ethtool -i, where
other drivers do.  The EEPROM version is located at offset 0x29.  This
patch adds support to display it.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoixgbe: Update link setup code to better support autonegotiation of speed
Peter P Waskiewicz Jr [Fri, 6 Feb 2009 07:54:21 +0000 (23:54 -0800)]
ixgbe: Update link setup code to better support autonegotiation of speed

The current code has some flaws in it when performing autonegotiation,
especially on KX/KX4 links.  This patch updates the code to better handle
the autonegotiation states on link setup.  The patch also removes a redundant
link configuration call on driver load, and moves link configuration to
the ->open() path.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoixgbe: Refactor set_num_queues() and cache_ring_register()
Peter P Waskiewicz Jr [Fri, 6 Feb 2009 07:53:59 +0000 (23:53 -0800)]
ixgbe: Refactor set_num_queues() and cache_ring_register()

The current code to determine the number of queues the device will want
on driver initialization is ugly and difficult to maintain.  It also
doesn't allow for easy expansion for future features or future hardware.
This patch refactors these routines, and make them easier to deal with.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogro: Fix frag_list merging on imprecisely split packets
Herbert Xu [Fri, 6 Feb 2009 05:26:52 +0000 (21:26 -0800)]
gro: Fix frag_list merging on imprecisely split packets

The previous fix ad0f9904444de1309dedd2b9e365cae8af77d9b1 (gro:
Fix handling of imprecisely split packets) only fixed the case
of frags merging, frag_list merging in the same circumstances
were still broken.

In particular, the packet headers end up in the data stream.

This patch fixes this plus another issue where an imprecisely
split packet header may be read incorrectly (this is mostly
harmless since it'll simply cause the packet to not match and
be rejected for GRO).

Thanks to Emil Tantilov and Jeff Kirsher for helping to track
this down.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: fix some trailing whitespaces
Graf Yang [Fri, 6 Feb 2009 05:26:19 +0000 (21:26 -0800)]
net: fix some trailing whitespaces

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agotun: Limit amount of queued packets per device
Herbert Xu [Fri, 6 Feb 2009 05:25:32 +0000 (21:25 -0800)]
tun: Limit amount of queued packets per device

Unlike a normal socket path, the tuntap device send path does
not have any accounting.  This means that the user-space sender
may be able to pin down arbitrary amounts of kernel memory by
continuing to send data to an end-point that is congested.

Even when this isn't an issue because of limited queueing at
most end points, this can also be a problem because its only
response to congestion is packet loss.  That is, when those
local queues at the end-point fills up, the tuntap device will
start wasting system time because it will continue to send
data there which simply gets dropped straight away.

Of course one could argue that everybody should do congestion
control end-to-end, unfortunately there are people in this world
still hooked on UDP, and they don't appear to be going away
anywhere fast.  In fact, we've always helped them by performing
accounting in our UDP code, the sole purpose of which is to
provide congestion feedback other than through packet loss.

This patch attempts to apply the same bandaid to the tuntap device.
It creates a pseudo-socket object which is used to account our
packets just as a normal socket does for UDP.  Of course things
are a little complex because we're actually reinjecting traffic
back into the stack rather than out of the stack.

The stack complexities however should have been resolved by preceding
patches.  So this one can simply start using skb_set_owner_w.

For now the accounting is essentially disabled by default for
backwards compatibility.  In particular, we set the cap to INT_MAX.
This is so that existing applications don't get confused by the
sudden arrival EAGAIN errors.

In future we may wish (or be forced to) do this by default.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoRevert "tcp: Always set urgent pointer if it's beyond snd_nxt"
David S. Miller [Thu, 5 Feb 2009 23:38:31 +0000 (15:38 -0800)]
Revert "tcp: Always set urgent pointer if it's beyond snd_nxt"

This reverts commit 64ff3b938ec6782e6585a83d5459b98b0c3f6eb8.

Jeff Chua reports that it breaks rlogin for him.

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6: Copy cork options in ip6_append_data
Herbert Xu [Thu, 5 Feb 2009 23:15:50 +0000 (15:15 -0800)]
ipv6: Copy cork options in ip6_append_data

As the options passed to ip6_append_data may be ephemeral, we need
to duplicate it for corking.  This patch applies the simplest fix
which is to memdup all the relevant bits.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Thu, 5 Feb 2009 23:08:11 +0000 (15:08 -0800)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

15 years agoudp: Fix UDP short packet false positive
Jesper Dangaard Brouer [Thu, 5 Feb 2009 23:05:45 +0000 (15:05 -0800)]
udp: Fix UDP short packet false positive

The UDP header pointer assignment must happen after calling
pskb_may_pull().  As pskb_may_pull() can potentially alter the SKB
buffer.

This was exposted by running multicast traffic through the NIU driver,
as it won't prepull the protocol headers into the linear area on
receive.

Signed-off-by: Jesper Dangaard Brouer <hawk@comx.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Reexport sock_alloc_send_pskb
Herbert Xu [Thu, 5 Feb 2009 00:55:54 +0000 (16:55 -0800)]
net: Reexport sock_alloc_send_pskb

The function sock_alloc_send_pskb is completely useless if not
exported since most of the code in it won't be used as is.  In
fact, this code has already been duplicated in the tun driver.

Now that we need accounting in the tun driver, we can in fact
use this function as is.  So this patch marks it for export again.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Partially allow skb destructors to be used on receive path
Herbert Xu [Thu, 5 Feb 2009 00:55:27 +0000 (16:55 -0800)]
net: Partially allow skb destructors to be used on receive path

As it currently stands, skb destructors are forbidden on the
receive path because the protocol end-points will overwrite
any existing destructor with their own.

This is the reason why we have to call skb_orphan in the loopback
driver before we reinject the packet back into the stack, thus
creating a period during which loopback traffic isn't charged
to any socket.

With virtualisation, we have a similar problem in that traffic
is reinjected into the stack without being associated with any
socket entity, thus providing no natural congestion push-back
for those poor folks still stuck with UDP.

Now had we been consistent in telling them that UDP simply has
no congestion feedback, I could just fob them off.  Unfortunately,
we appear to have gone to some length in catering for this on
the standard UDP path, with skb/socket accounting so that has
created a very unhealthy dependency.

Alas habits are difficult to break out of, so we may just have
to allow skb destructors on the receive path.

It turns out that making skb destructors useable on the receive path
isn't as easy as it seems.  For instance, simply adding skb_orphan
to skb_set_owner_r isn't enough.  This is because we assume all
over the IP stack that skb->sk is an IP socket if present.

The new transparent proxy code goes one step further and assumes
that skb->sk is the receiving socket if present.

Now all of this can be dealt with by adding simple checks such
as only treating skb->sk as an IP socket if skb->sk->sk_family
matches.  However, it turns out that for bridging at least we
don't need to do all of this work.

This is of interest because most virtualisation setups use bridging
so we don't actually go through the IP stack on the host (with
the exception of our old nemesis the bridge netfilter, but that's
easily taken care of).

So this patch simply adds skb_orphan to the point just before we
enter the IP stack, but after we've gone through the bridge on the
receive path.  It also adds an skb_orphan to the one place in
netfilter that touches skb->sk/skb->destructor, that is, tproxy.

One word of caution, because of the internal code structure, anyone
wishing to deploy this must use skb_set_owner_w as opposed to
skb_set_owner_r since many functions that create a new skb from
an existing one will invoke skb_set_owner_w on the new skb.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Thu, 5 Feb 2009 00:52:41 +0000 (16:52 -0800)]
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

15 years agoMerge branch 'master' of /home/davem/src/GIT/linux-2.6/
David S. Miller [Thu, 5 Feb 2009 00:51:58 +0000 (16:51 -0800)]
Merge branch 'master' of /home/davem/src/GIT/linux-2.6/

15 years agogianfar: Fix stashing support
Andy Fleming [Thu, 5 Feb 2009 00:43:44 +0000 (16:43 -0800)]
gianfar: Fix stashing support

Stashing is only supported on the 85xx (e500-based) SoCs.  The 83xx and 86xx
chips don't have a proper cache for this.  U-Boot has been updated to add
stashing properties to the device tree nodes of gianfar devices on 85xx.  So
now we modify Linux to keep stashing off unless those properties are there.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogianfar: Add support for skb recycling
Andy Fleming [Thu, 5 Feb 2009 00:43:16 +0000 (16:43 -0800)]
gianfar: Add support for skb recycling

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonetdev: Merge UCC and gianfar MDIO bus drivers
Andy Fleming [Thu, 5 Feb 2009 00:42:12 +0000 (16:42 -0800)]
netdev: Merge UCC and gianfar MDIO bus drivers

The MDIO bus drivers for the UCC and gianfar ethernet controllers are
essentially the same.  There's no reason to duplicate that much code.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogianfar: Fix potential soft reset race
Andy Fleming [Thu, 5 Feb 2009 00:38:05 +0000 (16:38 -0800)]
gianfar: Fix potential soft reset race

SOFT_RESET must be asserted for at least 3 TX clocks in order for it to work
properly.  The syncs in the gfar_write() commands have been hiding this, but
we need to guarantee it.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogianfar: Fix BD_LENGTH_MASK definition
Andy Fleming [Thu, 5 Feb 2009 00:37:40 +0000 (16:37 -0800)]
gianfar: Fix BD_LENGTH_MASK definition

BD_LENGTH_MASK is supposed to catch the low 16-bits of the status field, not
the low byte.  The old way, we would never be able to clean up tx packets with
sizes divisible by 256.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovirtio_net: Allow setting the MAC address of the NIC
Alex Williamson [Thu, 5 Feb 2009 00:36:34 +0000 (16:36 -0800)]
virtio_net: Allow setting the MAC address of the NIC

Many physical NICs let the OS re-program the "hardware" MAC
address.  Virtual NICs should allow this too.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Acked-by: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovirtio_net: Add support for VLAN filtering in the hypervisor
Alex Williamson [Wed, 4 Feb 2009 09:02:50 +0000 (09:02 +0000)]
virtio_net: Add support for VLAN filtering in the hypervisor

VLAN filtering allows the hypervisor to drop packets from VLANs
that we're not a part of, further reducing the number of extraneous
packets recieved.  This makes use of the VLAN virtqueue command class.
The CTRL_VLAN feature bit tells us whether the backend supports VLAN
filtering.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovirtio_net: Add a MAC filter table
Alex Williamson [Wed, 4 Feb 2009 09:02:45 +0000 (09:02 +0000)]
virtio_net: Add a MAC filter table

Make use of the MAC control virtqueue class to support a MAC
filter table.  The filter table is managed by the hypervisor.
We consider the table to be available if the CTRL_RX feature
bit is set.  We leave it to the hypervisor to manage the table
and enable promiscuous or all-multi mode as necessary depending
on the resources available to it.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovirtio_net: Add a set_rx_mode interface
Alex Williamson [Wed, 4 Feb 2009 09:02:40 +0000 (09:02 +0000)]
virtio_net: Add a set_rx_mode interface

Make use of the RX_MODE control virtqueue class to enable the
set_rx_mode netdev interface.  This allows us to selectively
enable/disable promiscuous and allmulti mode so we don't see
packets we don't want.  For now, we automatically enable these
as needed if additional unicast or multicast addresses are
requested.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovirtio_net: Add a virtqueue for outbound control commands
Alex Williamson [Wed, 4 Feb 2009 09:02:34 +0000 (09:02 +0000)]
virtio_net: Add a virtqueue for outbound control commands

This will be used for RX mode, MAC filter table, VLAN filtering, etc...

The control transaction consists of one or more "out" sg entries and
one or more "in" sg entries.  The first out entry contains a header
defining the class and command.  Additional out entries may provide
data for the command.  The last in entry provides a status response
back from the command.

Virtqueues typically run asynchronous, running a callback function
when there's data in the channel.  We can't readily make use of this
in the command paths where we need to use this.  Instead, we kick
the virtqueue and spin.  The kick causes an I/O write, triggering an
immediate trap into the hypervisor.

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agocxgb3: Fix lro switch
Divy Le Ray [Thu, 5 Feb 2009 00:31:39 +0000 (16:31 -0800)]
cxgb3: Fix lro switch

The LRO switch is always set to 1 in the rx processing loop.
It breaks the accelerated iSCSI receive traffic.
Fix its computation.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoiwlwifi: save PCI state before suspend, restore after resume
Reinette Chatre [Tue, 3 Feb 2009 18:20:03 +0000 (10:20 -0800)]
iwlwifi: save PCI state before suspend, restore after resume

This is the right thing to do and fixes the following warning:

[  115.012278] ------------[ cut here ]------------
[  115.012281] WARNING: at drivers/pci/pci-driver.c:370
pci_legacy_suspend+0x85/0xc2()
[  115.012285] Hardware name: Latitude D630
[  115.012301] PCI PM: Device state not saved by
iwl3945_pci_suspend+0x0/0x4c [iwl3945]
[  115.012304] Modules linked in: fuse nfsd lockd nfs_acl auth_rpcgss
exportfs sunrpc ipv6 acpi_cpufreq kvm_intel kvm snd_hda_codec_idt
snd_hda_intel snd_hda_codec snd_hwdep arc4 snd_seq_device snd_pcm_oss
snd_mixer_oss ecb snd_pcm cryptomgr aead snd_timer crypto_blkcipher
snd snd_page_alloc ohci1394 crypto_hash crypto_algapi ch341 ieee1394
usbserial thermal iwl3945 mac80211 led_class lib80211 tg3 processor
i2c_i801 i2c_core sg cfg80211 libphy usbhid battery ac button sr_mod
cdrom evdev dcdbas ata_generic ata_piix libata sd_mod scsi_mod ext3
jbd mbcache uhci_hcd ohci_hcd ehci_hcd usbcore [last unloaded:
microcode]
[  115.012374] Pid: 4163, comm: pm-suspend Not tainted
2.6.29-rc3-00227-gf1dd849-dirty #67
[  115.012377] Call Trace:
[  115.012382]  [<ffffffff8023d04d>] warn_slowpath+0xb1/0xed
[  115.012387]  [<ffffffff80450b5e>] ? _spin_unlock_irqrestore+0x5c/0x78
[  115.012390]  [<ffffffff80254f08>] ? up+0x34/0x39
[  115.012394]  [<ffffffff80362319>] ? acpi_ut_release_mutex+0x5d/0x61
[  115.012397]  [<ffffffff803584b2>] ? acpi_get_data+0x5e/0x70
[  115.012400]  [<ffffffff80363dd9>] ? acpi_bus_get_device+0x25/0x39
[  115.012403]  [<ffffffff80363e98>] ? acpi_bus_power_manageable+0x11/0x29
[  115.012406]  [<ffffffff803462f7>] ? acpi_pci_power_manageable+0x17/0x19
[  115.012410]  [<ffffffff8033ddfd>] ? pci_set_power_state+0xcc/0x101
[  115.012418]  [<ffffffffa01f28e9>] ? iwl3945_pci_suspend+0x0/0x4c [iwl3945]
[  115.012422]  [<ffffffff803401e6>] pci_legacy_suspend+0x85/0xc2
[  115.012425]  [<ffffffff80340316>] pci_pm_suspend+0x34/0x86
[  115.012429]  [<ffffffff8039d7ce>] pm_op+0x52/0xe5
[  115.012432]  [<ffffffff8039dd78>] device_suspend+0x32a/0x451
[  115.012436]  [<ffffffff80269ec2>] suspend_devices_and_enter+0x3e/0x13a
[  115.012439]  [<ffffffff8026a128>] enter_state+0x110/0x164
[  115.012442]  [<ffffffff8026a233>] state_store+0xb7/0xd7
[  115.012446]  [<ffffffff8032f95f>] kobj_attr_store+0x17/0x19
[  115.012449]  [<ffffffff80307d64>] sysfs_write_file+0xe4/0x119
[  115.012453]  [<ffffffff802baa7a>] vfs_write+0xae/0x137
[  115.012456]  [<ffffffff802babc7>] sys_write+0x47/0x70
[  115.012459]  [<ffffffff8020b73a>] system_call_fastpath+0x16/0x1b
[  115.012467] ---[ end trace 829828966f6f24dc ]---

Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Tested-by: Ming Lei <tom.leiming@gmail.com>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoiwlwifi: clean key table in iwl_clear_stations_table
Reinette Chatre [Wed, 28 Jan 2009 17:38:30 +0000 (09:38 -0800)]
iwlwifi: clean key table in iwl_clear_stations_table

Cleans uCode key table bit map iwl_clear_stations_table
since all stations are cleared also the key table must be.

Since the keys are not removed properly on suspend by mac80211
this may result in exhausting key table on resume leading
to memory corruption during removal

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney...
Linus Torvalds [Wed, 4 Feb 2009 15:56:25 +0000 (07:56 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/cooloney/blackfin-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/blackfin-2.6: (40 commits)
  Blackfin arch: Remove outdated code
  Blackfin arch: Fix udelay implementation
  Blackfin arch: Update Copyright information
  Blackfin arch: Add BF561 PPI POLS, POLC Masks
  Blackfin arch: Update CM-BF527 kernel config
  Blackfin arch: define bfin_memmap as static since it is only used here
  Blackfin arch: cplb mananger: use a do...while loop rather than a for loop
  Blackfin arch: fix bug - traps test case 19 for exception 0x2d fails
  Blackfin arch: add platform device bfin_mii-bus and KSZ8893M switch driver platform resources to board files
  Blackfin arch: build jtag tty driver as a module by default
  Blackfin arch: fix 2 bugs related to debug
  Blackfin arch: Add ANOMALY_05000380 to BF54x to kill the compile warning
  Blackfin arch: Fix bug - 561 SMP kernel can't boot from jffs2
  Blackfin arch: base SIC_IWR# programming on whether the MMR exists
  Blackfin arch: read SYSCR on newer parts that mirror the bits of SWRST in it
  Blackfin arch: fixup board init function name
  Blackfin arch: drop CONFIG_I2C_BOARDINFO ifdefs
  Blackfin arch: bfin_reset->_bfin_reset redirection no longer needed
  Blackfin arch: sync reboot handler with version in u-boot
  Blackfin arch: Faster Implementation of csum_tcpudp_nofold()
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Wed, 4 Feb 2009 15:54:00 +0000 (07:54 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc64: Kill bogus TPC/address truncation during 32-bit faults.
  sparc: fixup for sparseirq changes
  sparc64: Validate kernel generated fault addresses on sparc64.
  sparc64: On non-Niagara, need to touch NMI watchdog in NOHZ mode.
  sparc64: Implement NMI watchdog on capable cpus.
  sparc: Probe PMU type and record in sparc_pmu_type.
  sparc64: Move generic PCR support code to seperate file.

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Wed, 4 Feb 2009 15:52:21 +0000 (07:52 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  sunrpc: fix rdma dependencies
  e1000: Fix PCI enable to honor the need_ioport flag
  sgi-xp: link XPNET's net_device_ops to its net_device structure
  pcnet_cs: Fix misuse of the equality operator.
  hso: add new device id's
  dca: redesign locks to fix deadlocks
  cassini/sungem: limit reaches -1, but 0 tested
  net: variables reach -1, but 0 tested
  qlge: bugfix: Add missing netif_napi_del call.
  qlge: bugfix: Add flash offset for second port.
  qlge: bugfix: Fix endian issue when reading flash.
  udp: increments sk_drops in __udp_queue_rcv_skb()
  net: Fix userland breakage wrt. linux/if_tunnel.h
  net: packet socket packet_lookup_frame fix

15 years agoMerge branch 'for-linus' of git://git.o-hand.com/linux-mfd
Linus Torvalds [Wed, 4 Feb 2009 15:40:54 +0000 (07:40 -0800)]
Merge branch 'for-linus' of git://git.o-hand.com/linux-mfd

* 'for-linus' of git://git.o-hand.com/linux-mfd:
  mfd: Remove non exported references from pcf50633

15 years agoBlackfin arch: Remove outdated code
Michael Hennerich [Wed, 4 Feb 2009 08:49:45 +0000 (16:49 +0800)]
Blackfin arch: Remove outdated code

The removed version with the loop registers saved on the stack was
originally intended to workaround the missing toolchain support for
LoopReg Clobbers.

Since our toolchain now supports these there is no point in keeping this
workaround. And since we don't touch LoopRegs anymore we're no longer
subject for ANOMALY_05000312.

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
15 years agoBlackfin arch: Fix udelay implementation
Michael Hennerich [Wed, 4 Feb 2009 08:49:45 +0000 (16:49 +0800)]
Blackfin arch: Fix udelay implementation

Avoid possible overflow during 32*32->32 multiplies.

Reported-by: Marco Reppenhagen <marco.reppenhagen@auerswald.de>
Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>