openwrt/staging/blogic.git
13 years agonet: Add ethtool to mii advertisment conversion helpers
Matt Carlson [Wed, 16 Nov 2011 23:36:59 +0000 (18:36 -0500)]
net: Add ethtool to mii advertisment conversion helpers

Translating between ethtool advertisement settings and MII
advertisements are common operations for ethernet drivers.  This patch
adds a set of helper functions that implements the conversion.  The
patch then modifies a couple of the drivers to use the new functions.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net...
David S. Miller [Wed, 16 Nov 2011 23:31:56 +0000 (18:31 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net-next

13 years agoteam: replicate options on register
Jiri Pirko [Wed, 16 Nov 2011 11:09:09 +0000 (11:09 +0000)]
team: replicate options on register

Since multiple team instances are putting defined options into their
option list, during register each option must be cloned before added
into list. This resolves uncool memory corruptions when using multiple
teams.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoteam: convert overall spinlock to mutex
Jiri Pirko [Wed, 16 Nov 2011 11:09:08 +0000 (11:09 +0000)]
team: convert overall spinlock to mutex

No need to have spinlock for this purpose. So convert this to mutex and
avoid current schedule while atomic problems in netlink code.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoteam: Do not hold rcu_read_lock when running netlink cmds
Jiri Pirko [Wed, 16 Nov 2011 11:55:54 +0000 (11:55 +0000)]
team: Do not hold rcu_read_lock when running netlink cmds

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: double free in lowpan_fragment_xmit()
Dan Carpenter [Tue, 15 Nov 2011 22:36:43 +0000 (22:36 +0000)]
6LoWPAN: double free in lowpan_fragment_xmit()

dev_queue_xmit() consumes its own skb, so the call to dev_kfree_skb()
in lowpan_fragment_xmit() is a double free.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: Use V1 query link status command for lancer
Padmanabh Ratnakar [Wed, 16 Nov 2011 02:03:45 +0000 (02:03 +0000)]
be2net: Use V1 query link status command for lancer

Use V1 version of query link status command for Lancer.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: Fix VLAN promiscous mode for Lancer
Padmanabh Ratnakar [Wed, 16 Nov 2011 02:03:32 +0000 (02:03 +0000)]
be2net: Fix VLAN promiscous mode for Lancer

To enable VLAN promiscous mode, the HW interface should be created
with VLAN promiscous capability in Lancer. Add this capability during
creation of the HW interface.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: Add EEPROM dump feature for Lancer
Padmanabh Ratnakar [Wed, 16 Nov 2011 02:03:07 +0000 (02:03 +0000)]
be2net: Add EEPROM dump feature for Lancer

Implemented eeprom dump using ethtool feature for Lancer.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: add register dump feature for Lancer
Padmanabh Ratnakar [Wed, 16 Nov 2011 02:02:43 +0000 (02:02 +0000)]
be2net: add register dump feature for Lancer

Implement register dump using ethtool for Lancer.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: Fix TX queue create for Lancer
Padmanabh Ratnakar [Wed, 16 Nov 2011 02:02:23 +0000 (02:02 +0000)]
be2net: Fix TX queue create for Lancer

Lancer uses V1 version of TXQ create. This command needs interface
id for TX queue creation. Rearrange code such that tx queue create
is after interface create. As TXQ create is now called after MCC
ring create use MCC instead of MBOX.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoinfiniband: Update net drivers for netdev_features_t changes.
David S. Miller [Wed, 16 Nov 2011 23:05:50 +0000 (18:05 -0500)]
infiniband: Update net drivers for netdev_features_t changes.

Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: remove NETIF_F_NO_CSUM feature bit
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: remove NETIF_F_NO_CSUM feature bit

Only distinct use is checking if NETIF_F_NOCACHE_COPY should be
enabled by default. The check heuristics is altered a bit here,
so it hits other people than before. The default shouldn't be
trusted for performance-critical cases anyway.

For all other uses NETIF_F_NO_CSUM is equivalent to NETIF_F_HW_CSUM.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: extend netdev_features_t to 64 bits
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: extend netdev_features_t to 64 bits

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoethtool: prepare for larger netdev_features_t type
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
ethtool: prepare for larger netdev_features_t type

v2: changed loop in ethtool_set_features() per Ben's suggestion

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: ethtool: use C99 array initialization for feature-names table
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: ethtool: use C99 array initialization for feature-names table

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: Define enum for net device features.
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: Define enum for net device features.

Define feature values by bit position instead of direct 2**i values
and force the values to be of type netdev_features_t.

Cleaned and extended from patch by Mahesh Bandewar <maheshb@google.com>:
+ added netdev_features_t casts
+ included bits under NETIF_F_GSO_MASK
+ moved feature #defines out of struct net_device definition

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: introduce and use netdev_features_t for device features sets
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: introduce and use netdev_features_t for device features sets

v2: add couple missing conversions in drivers
split unexporting netdev_fix_features()
implemented %pNF
convert sock::sk_route_(no?)caps

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: split netdev features to separate header
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: split netdev features to separate header

Move features definitions to separate header so that linux/skbuff.h won't
need to include linux/netdevice.h after netdev_features_t is introduced.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: ethtool: break association of ETH_FLAG_* with NETIF_F_*
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: ethtool: break association of ETH_FLAG_* with NETIF_F_*

This is the only place left where dev->features are directly
exposed to userspace.

I know checkpatch.pl complains about __ethtool_{get,set}_flags(), but
the code is easier to read this way.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: remove legacy ethtool ops
Michał Mirosław [Tue, 15 Nov 2011 15:29:55 +0000 (15:29 +0000)]
net: remove legacy ethtool ops

As all drivers are converted, we may now remove discrete offload setting
callback handling.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: sweep-up some straglers in strlcpy conversion of .get_drvinfo routines
Rick Jones [Tue, 15 Nov 2011 14:59:53 +0000 (14:59 +0000)]
net: sweep-up some straglers in strlcpy conversion of .get_drvinfo routines

Convert some remaining straglers' .get_drvinfo routines to use strlcpy
rather than strcpy/strncpy.

Signed-off-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: use jump_label for netstamp_needed
Eric Dumazet [Tue, 15 Nov 2011 04:12:55 +0000 (04:12 +0000)]
net: use jump_label for netstamp_needed

netstamp_needed seems a good candidate to jump_label conversion.

This avoids 3 conditional branches per incoming packet in fast path.

No measurable difference, given that these conditional branches are
predicted on modern cpus. Only a small icache reduction, thanks to the
unlikely() stuff.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenable virtio_net to return bus_info in ethtool -i consistent with emulated NICs
Rick Jones [Mon, 14 Nov 2011 14:17:08 +0000 (14:17 +0000)]
enable virtio_net to return bus_info in ethtool -i consistent with emulated NICs

Add a new .bus_name to virtio_config_ops then modify virtio_net to
call through to it in an ethtool .get_drvinfo routine to report
bus_info in ethtool -i output which is consistent with other
emulated NICs and the output of lspci.

Signed-off-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agointel: Convert <FOO>_LENGTH_OF_ADDRESS to ETH_ALEN
Joe Perches [Tue, 20 Sep 2011 15:32:52 +0000 (15:32 +0000)]
intel: Convert <FOO>_LENGTH_OF_ADDRESS to ETH_ALEN

Use the normal #defines not module specific ones.

Signed-off-by: Joe Perches <joe@perches.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoigb: Convert bare printk to pr_notice
Joe Perches [Fri, 21 Oct 2011 20:04:09 +0000 (20:04 +0000)]
igb: Convert bare printk to pr_notice

printks should use KERN_ levels.

Signed-off-by: Joe Perches <joe@perches.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoigb: Convert printks to pr_<level>
Jeff Kirsher [Fri, 21 Oct 2011 20:01:34 +0000 (20:01 +0000)]
igb: Convert printks to pr_<level>

Based on original patch from Joe Perches <joe@perches.com>

Use the current logging styles.

pr_<level> conversions are now prefixed with "igb: "

Correct a defect where the trailing NTU may have been printed
on a separate line because of an interleaving hex_dump.

Remove unnecessary uses of KERN_CONT and use single pr_info()s
to avoid any possible output interleaving from other modules.

Coalesce formats as appropriate.

-v2 fix-up to make checkpatch.pl compliant and remove change to
    the copyright line

CC: Joe Perches <joe@perches.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
13 years agoixgbevf: Convert printks to pr_<level>
Jeff Kirsher [Fri, 21 Oct 2011 19:38:18 +0000 (19:38 +0000)]
ixgbevf: Convert printks to pr_<level>

Based on the original patch from Joe Perches <joe@perches.com>

Use the current logging styles, prefix output with "ixgbevf: "

Add #define pr_fmt
Coalesce formats.

-v2 Fix-up to make checkpatch.pl compliant and remove change to
    copyright line

CC: Joe Perches <joe@perches.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
13 years agoigbvf: Convert printks to pr_<level>
Jeff Kirsher [Fri, 21 Oct 2011 19:42:26 +0000 (19:42 +0000)]
igbvf: Convert printks to pr_<level>

Based on the previous patch from Joe Perches <joe@perches.com>

Use current logging styles.

Prefix all output via #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Neaten link status dev_info.

-v2 Remove Copyright changes and fix-up patch to make it checkpatch.pl
    compliant.

CC: Joe Perches <joe@perches.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
13 years agoixgbe: fix LED blink logic to check for link
Emil Tantilov [Sat, 29 Oct 2011 06:54:55 +0000 (06:54 +0000)]
ixgbe: fix LED blink logic to check for link

Previously the driver would force link without checking whether the link was
already established. This caused some inconsistencies in the LED blink rate.

Do not force link if link is already up.

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>
13 years agoe1000e: Convert printks to pr_<level>
Jeff Kirsher [Thu, 3 Nov 2011 11:40:28 +0000 (11:40 +0000)]
e1000e: Convert printks to pr_<level>

Based on the original patch from Joe Perches.
Use the current logging styles.
pr_<level> conversions are now prefixed with "e1000e:"
Correct a couple of defects where the trailing NTU may have
been printed on a separate line because of an interleaving
hex_dump.
Remove unnecessary uses of KERN_CONT and use single pr_info()s
to avoid any possible output interleaving from other modules.
Coalesce formats as appropriate.
Remove an extra space from a broken across lines
coalescing of "Link Status " and " Change".

-v2 Remove changes to Copyright string

CC: Joe Perches <joe@perches.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
13 years agoe1000e: convert to real ndo_set_rx_mode
Jesse Brandeburg [Fri, 4 Nov 2011 05:47:06 +0000 (05:47 +0000)]
e1000e: convert to real ndo_set_rx_mode

Commit afc4b13d (net: remove use of ndo_set_multicast_list in
drivers) changed e1000e to use the ndo_set_rx_mode entry point,
but didn't implement the unicast address programming
functionality. Implement it to achieve the ability to add unicast
addresses.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agomdio-gpio: Add reset functionality to mdio-gpio driver(v2).
Srinivas Kandagatla [Tue, 15 Nov 2011 11:54:15 +0000 (11:54 +0000)]
mdio-gpio: Add reset functionality to mdio-gpio driver(v2).

This patch adds phy reset functionality to mdio-gpio driver. Now
mdio_gpio_platform_data has new member as function pointer which can be
filled at the bsp level for a callback from phy infrastructure. Also the
mdio-bitbang driver fills-in the reset function of mii_bus structure.

Without this patch the bsp level code has to takecare of the reseting
PHY's on the bus, which become bit hacky for every bsp and
phy-infrastructure is ignored aswell.

Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoIPv6: Removing unnecessary NULL checks.
Matti Vaittinen [Tue, 15 Nov 2011 00:58:59 +0000 (00:58 +0000)]
IPv6: Removing unnecessary NULL checks.

This patch removes unnecessary NULL checks noticed by Dan Carpenter.
Checks were introduced in commit
4a287eba2de395713d8b2b2aeaa69fa086832d34 to net-next.

Signed-off-by: Matti Vaittinen <Mazziesaccount@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv4: fix a memory leak in ic_bootp_send_if
RongQing.Li [Mon, 14 Nov 2011 19:37:24 +0000 (14:37 -0500)]
ipv4: fix a memory leak in ic_bootp_send_if

when dev_hard_header() failed, the newly allocated skb should be freed.

Signed-off-by: RongQing.Li <roy.qing.li@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: add endline at end of message
Dmitry Kravkov [Mon, 14 Nov 2011 19:36:40 +0000 (14:36 -0500)]
bnx2x: add endline at end of message

Reported-by: Joe Perches <joe@perches.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoIPv6 routing, NLM_F_* flag support: REPLACE and EXCL flags support, warn about missin...
Matti Vaittinen [Mon, 14 Nov 2011 00:15:14 +0000 (00:15 +0000)]
IPv6 routing, NLM_F_* flag support: REPLACE and EXCL flags support, warn about missing CREATE flag

The support for NLM_F_* flags at IPv6 routing requests.

If NLM_F_CREATE flag is not defined for RTM_NEWROUTE request,
warning is printed, but no error is returned. Instead new route is
added. Later NLM_F_CREATE may be required for
new route creation.

Exception is when NLM_F_REPLACE flag is given without NLM_F_CREATE, and
no matching route is found. In this case it should be safe to assume
that the request issuer is familiar with NLM_F_* flags, and does really
not want route to be created.

Specifying NLM_F_REPLACE flag will now make the kernel to search for
matching route, and replace it with new one. If no route is found and
NLM_F_CREATE is specified as well, then new route is created.

Also, specifying NLM_F_EXCL will yield returning of error if matching
route is found.

Patch created against linux-3.2-rc1

Signed-off-by: Matti Vaittinen <Mazziesaccount@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoIPv6 routing, NLM_F_* flag support: warn if new route is created without NLM_F_CREATE
Matti Vaittinen [Mon, 14 Nov 2011 00:14:49 +0000 (00:14 +0000)]
IPv6 routing, NLM_F_* flag support: warn if new route is created without NLM_F_CREATE

The support for NLM_F_* flags at IPv6 routing requests.

Warn if NLM_F_CREATE flag is not defined for RTM_NEWROUTE request,
creating new table. Later NLM_F_CREATE may be required for
new route creation.

Patch created against linux-3.2-rc1

Signed-off-by: Matti Vaittinen <Mazziesaccount@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet/can/mscan: Fix buggy listen only mode setting
Wolfgang Grandegger [Mon, 14 Nov 2011 19:30:05 +0000 (14:30 -0500)]
net/can/mscan: Fix buggy listen only mode setting

This patch fixes an issue introduced recently with commit
452448f9283e1939408b397e87974a418825b0a8.

CC: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoSweep the last of the active .get_drvinfo floors under ethernet/
Rick Jones [Mon, 14 Nov 2011 08:13:25 +0000 (08:13 +0000)]
Sweep the last of the active .get_drvinfo floors under ethernet/

This round of floor sweeping converts strncpy calls in various .get_drvinfo
routines to the preferred strlcpy.  It also does a modicum of other
cleaning in those routines.

Signed-off-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: uses build_skb() in receive path
Eric Dumazet [Mon, 14 Nov 2011 06:05:34 +0000 (06:05 +0000)]
bnx2x: uses build_skb() in receive path

bnx2x uses following formula to compute its rx_buf_sz :

dev->mtu + 2*L1_CACHE_BYTES + 14 + 8 + 8 + 2

Then core network adds NET_SKB_PAD and SKB_DATA_ALIGN(sizeof(struct
skb_shared_info))

Final allocated size for skb head on x86_64 (L1_CACHE_BYTES = 64,
MTU=1500) : 2112 bytes : SLUB/SLAB round this to 4096 bytes.

Since skb truesize is then bigger than SK_MEM_QUANTUM, we have lot of
false sharing because of mem_reclaim in UDP stack.

One possible way to half truesize is to reduce the need by 64 bytes
(2112 -> 2048 bytes)

Instead of allocating a full cache line at the end of packet for
alignment, we can use the fact that skb_shared_info sits at the end of
skb->head, and we can use this room, if we convert bnx2x to new
build_skb() infrastructure.

skb_shared_info will be initialized after hardware finished its
transfert, so we can eventually overwrite the final padding.

Using build_skb() also reduces cache line misses in the driver, since we
use cache hot skb instead of cold ones. Number of in-flight sk_buff
structures is lower, they are recycled while still hot.

Performance results :

(820.000 pps on a rx UDP monothread benchmark, instead of 720.000 pps)

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: introduce build_skb()
Eric Dumazet [Mon, 14 Nov 2011 06:03:34 +0000 (06:03 +0000)]
net: introduce build_skb()

One of the thing we discussed during netdev 2011 conference was the idea
to change some network drivers to allocate/populate their skb at RX
completion time, right before feeding the skb to network stack.

In old days, we allocated skbs when populating the RX ring.

This means bringing into cpu cache sk_buff and skb_shared_info cache
lines (since we clear/initialize them), then 'queue' skb->data to NIC.

By the time NIC fills a frame in skb->data buffer and host can process
it, cpu probably threw away the cache lines from its caches, because lot
of things happened between the allocation and final use.

So the deal would be to allocate only the data buffer for the NIC to
populate its RX ring buffer. And use build_skb() at RX completion to
attach a data buffer (now filled with an ethernet frame) to a new skb,
initialize the skb_shared_info portion, and give the hot skb to network
stack.

build_skb() is the function to allocate an skb, caller providing the
data buffer that should be attached to it. Drivers are expected to call
skb_reserve() right after build_skb() to adjust skb->data to the
Ethernet frame (usually skipping NET_SKB_PAD and NET_IP_ALIGN, but some
drivers might add a hardware provided alignment)

Data provided to build_skb() MUST have been allocated by a prior
kmalloc() call, with enough room to add SKB_DATA_ALIGN(sizeof(struct
skb_shared_info)) bytes at the end of the data without corrupting
incoming frame.

data = kmalloc(NET_SKB_PAD + NET_IP_ALIGN + 1536 +
               SKB_DATA_ALIGN(sizeof(struct skb_shared_info)),
       GFP_ATOMIC);
...
skb = build_skb(data);
if (!skb) {
recycle_data(data);
} else {
skb_reserve(skb, NET_SKB_PAD + NET_IP_ALIGN);
...
}

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Ben Hutchings <bhutchings@solarflare.com>
CC: Tom Herbert <therbert@google.com>
CC: Jamal Hadi Salim <hadi@mojatatu.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
CC: Thomas Graf <tgraf@infradead.org>
CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: fsl_pq_mdio: fix non tbi phy access
Baruch Siach [Mon, 14 Nov 2011 06:21:30 +0000 (08:21 +0200)]
net: fsl_pq_mdio: fix non tbi phy access

Since 952c5ca1 (fsl_pq_mdio: Clean up tbi address configuration) .probe returns
-EBUSY when the "tbi-phy" node is missing. Fix this.

Cc: Andy Fleming <afleming@freescale.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet/can/mscan: add listen only mode
Marc Kleine-Budde [Wed, 9 Nov 2011 10:50:49 +0000 (10:50 +0000)]
net/can/mscan: add listen only mode

This patch adds listen only mode to the mscan controller.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoneigh: new unresolved queue limits
Eric Dumazet [Wed, 9 Nov 2011 12:07:14 +0000 (12:07 +0000)]
neigh: new unresolved queue limits

Le mercredi 09 novembre 2011 à 16:21 -0500, David Miller a écrit :
> From: David Miller <davem@davemloft.net>
> Date: Wed, 09 Nov 2011 16:16:44 -0500 (EST)
>
> > From: Eric Dumazet <eric.dumazet@gmail.com>
> > Date: Wed, 09 Nov 2011 12:14:09 +0100
> >
> >> unres_qlen is the number of frames we are able to queue per unresolved
> >> neighbour. Its default value (3) was never changed and is responsible
> >> for strange drops, especially if IP fragments are used, or multiple
> >> sessions start in parallel. Even a single tcp flow can hit this limit.
> >  ...
> >
> > Ok, I've applied this, let's see what happens :-)
>
> Early answer, build fails.
>
> Please test build this patch with DECNET enabled and resubmit.  The
> decnet neigh layer still refers to the removed ->queue_len member.
>
> Thanks.

Ouch, this was fixed on one machine yesterday, but not the other one I
used this morning, sorry.

[PATCH V5 net-next] neigh: new unresolved queue limits

unres_qlen is the number of frames we are able to queue per unresolved
neighbour. Its default value (3) was never changed and is responsible
for strange drops, especially if IP fragments are used, or multiple
sessions start in parallel. Even a single tcp flow can hit this limit.

$ arp -d 192.168.20.108 ; ping -c 2 -s 8000 192.168.20.108
PING 192.168.20.108 (192.168.20.108) 8000(8028) bytes of data.
8008 bytes from 192.168.20.108: icmp_seq=2 ttl=64 time=0.322 ms

Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobridge: add NTF_USE support
stephen hemminger [Wed, 9 Nov 2011 18:30:08 +0000 (18:30 +0000)]
bridge: add NTF_USE support

More changes to the recent code to support control of forwarding
database via netlink.
   * Support NTF_USE like neighbour table
   * Validate state bits from application
   * Only send notifications (and change bits) if new entry is
     different.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoSweep additional floors of strcpy in .get_drvinfo routines
Rick Jones [Wed, 9 Nov 2011 09:58:07 +0000 (09:58 +0000)]
Sweep additional floors of strcpy in .get_drvinfo routines

Perform another round of floor sweeping, converting the .get_drvinfo
routines of additional drivers from strcpy to strlcpy along with
some conversion of sprintf to snprintf.

Signed-off-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agofsl_pq_mdio: Clean up tbi address configuration
Andy Fleming [Fri, 11 Nov 2011 05:10:39 +0000 (05:10 +0000)]
fsl_pq_mdio: Clean up tbi address configuration

The code for setting the address of the internal TBI PHY was
convoluted enough without a maze of ifdefs. Clean it up a bit
so we allow the logic to fail down to -ENODEV at the end of
the if/else ladder, rather than using ifdefs to repeat the same
failure code over and over.

Also, remove the support for the auto-configuration. I'm not aware of
anyone using it, and it ends up using the bus mutex before it's been
initialized.

Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet-forcedeth: Add internal loopback support for forcedeth NICs.
Sanjay Hortikar [Fri, 11 Nov 2011 16:11:21 +0000 (16:11 +0000)]
net-forcedeth: Add internal loopback support for forcedeth NICs.

Support enabling/disabling/querying internal loopback mode for
forcedeth NICs using ethtool.

Signed-off-by: Sanjay Hortikar <horti@google.com>
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: update documentation
alex.bluesman.smirnov@gmail.com [Thu, 10 Nov 2011 07:41:11 +0000 (07:41 +0000)]
6LoWPAN: update documentation

This patch adds chapter to documentation which describes how to use
6lowpan technology.

Signed-off-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: UDP header decompression
alex.bluesman.smirnov@gmail.com [Thu, 10 Nov 2011 07:40:53 +0000 (07:40 +0000)]
6LoWPAN: UDP header decompression

This patch provides possibility to decompress UDP headers.
Derived from Contiki OS.

Signed-off-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: UDP header compression
alex.bluesman.smirnov@gmail.com [Thu, 10 Nov 2011 07:40:14 +0000 (07:40 +0000)]
6LoWPAN: UDP header compression

This patch adds support for UDP header compression.
Derived from Contiki OS.

Signed-off-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: set proper netdev flags
alex.bluesman.smirnov@gmail.com [Thu, 10 Nov 2011 07:39:37 +0000 (07:39 +0000)]
6LoWPAN: set proper netdev flags

This patch fixes settings for device initialization which makes possible to
use NDISC and TCP.

Signed-off-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Acked-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: disable debugging by default
alex.bluesman.smirnov@gmail.com [Thu, 10 Nov 2011 07:39:15 +0000 (07:39 +0000)]
6LoWPAN: disable debugging by default

This patch disables debug output enabled by default.

Signed-off-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Acked-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years ago6LoWPAN: add fragmentation support
alex.bluesman.smirnov@gmail.com [Thu, 10 Nov 2011 07:38:38 +0000 (07:38 +0000)]
6LoWPAN: add fragmentation support

This patch adds support for frame fragmentation.

Signed-off-by: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv6: reduce percpu needs for icmpv6msg mibs
Eric Dumazet [Sun, 13 Nov 2011 01:24:04 +0000 (01:24 +0000)]
ipv6: reduce percpu needs for icmpv6msg mibs

Reading /proc/net/snmp6 on a machine with a lot of cpus is very
expensive (can be ~88000 us).

This is because ICMPV6MSG MIB uses 4096 bytes per cpu, and folding
values for all possible cpus can read 16 Mbytes of memory (32MBytes on
non x86 arches)

ICMP messages are not considered as fast path on a typical server, and
eventually few cpus handle them anyway. We can afford an atomic
operation instead of using percpu data.

This saves 4096 bytes per cpu and per network namespace.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: introduce ethernet teaming device
Jiri Pirko [Fri, 11 Nov 2011 22:16:48 +0000 (22:16 +0000)]
net: introduce ethernet teaming device

This patch introduces new network device called team. It supposes to be
very fast, simple, userspace-driven alternative to existing bonding
driver.

Userspace library called libteam with couple of demo apps is available
here:
https://github.com/jpirko/libteam
Note it's still in its dipers atm.

team<->libteam use generic netlink for communication. That and rtnl
suppose to be the only way to configure team device, no sysfs etc.

Python binding of libteam was recently introduced.
Daemon providing arpmon/miimon active-backup functionality will be
introduced shortly. All what's necessary is already implemented in
kernel team driver.

v7->v8:
- check ndo_ndo_vlan_rx_[add/kill]_vid functions before calling
  them.
- use dev_kfree_skb_any() instead of dev_kfree_skb()

v6->v7:
- transmit and receive functions are not checked in hot paths.
  That also resolves memory leak on transmit when no port is
  present

v5->v6:
- changed couple of _rcu calls to non _rcu ones in non-readers

v4->v5:
- team_change_mtu() uses team->lock while travesing though port
  list
- mac address changes are moved completely to jurisdiction of
  userspace daemon. This way the daemon can do FOM1, FOM2 and
  possibly other weird things with mac addresses.
  Only round-robin mode sets up all ports to bond's address then
  enslaved.
- Extended Kconfig text

v3->v4:
- remove redundant synchronize_rcu from __team_change_mode()
- revert "set and clear of mode_ops happens per pointer, not per
  byte"
- extend comment of function __team_change_mode()

v2->v3:
- team_change_mtu() uses rcu version of list traversal to unwind
- set and clear of mode_ops happens per pointer, not per byte
- port hashlist changed to be embedded into team structure
- error branch in team_port_enter() does cleanup now
- fixed rtln->rtnl

v1->v2:
- modes are made as modules. Makes team more modular and
  extendable.
- several commenters' nitpicks found on v1 were fixed
- several other bugs were fixed.
- note I ignored Eric's comment about roundrobin port selector
  as Eric's way may be easily implemented as another mode (mode
  "random") in future.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: update driver version to 1.70.35-0
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:32 +0000 (04:34 +0000)]
bnx2x: update driver version to 1.70.35-0

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: Remove on-stack napi struct variable
Ariel Elior [Sun, 13 Nov 2011 04:34:31 +0000 (04:34 +0000)]
bnx2x: Remove on-stack napi struct variable

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: prevent race in statistics flow
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:30 +0000 (04:34 +0000)]
bnx2x: prevent race in statistics flow

The race may cause access of registers while MAC hw block is
in reset state. As a result syslog will show error messages.
We can prevent this by using state from local variable.

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: add fan failure event handling
Ariel Elior [Sun, 13 Nov 2011 04:34:29 +0000 (04:34 +0000)]
bnx2x: add fan failure event handling

Shut down the device in case of fan failure to prevent HW damage.

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: remove unused #define
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:28 +0000 (04:34 +0000)]
bnx2x: remove unused #define

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: simplify definition of RX_SGE_MASK_LEN and use it.
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:27 +0000 (04:34 +0000)]
bnx2x: simplify definition of RX_SGE_MASK_LEN and use it.

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: DCBX: use #define instead of magic
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:26 +0000 (04:34 +0000)]
bnx2x: DCBX: use #define instead of magic

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: propagate DCBX negotiation
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:25 +0000 (04:34 +0000)]
bnx2x: propagate DCBX negotiation

We need propagate the DCBX results from PMF to other functions
on the same port, in order to properly update netdev structure
and allow following new ETS and PFC configurations.

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: separate FCoE and iSCSI license initialization.
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:24 +0000 (04:34 +0000)]
bnx2x: separate FCoE and iSCSI license initialization.

FCoE license info must be initialized at probe(), but
iSCSI at open().

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: remove unused variable
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:23 +0000 (04:34 +0000)]
bnx2x: remove unused variable

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: use rx_queue index for skb_record_rx_queue()
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:22 +0000 (04:34 +0000)]
bnx2x: use rx_queue index for skb_record_rx_queue()

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2x: allow FCoE and DCB for 578xx
Dmitry Kravkov [Sun, 13 Nov 2011 04:34:21 +0000 (04:34 +0000)]
bnx2x: allow FCoE and DCB for 578xx

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: stop issuing FW cmds if any cmd times out
Sathya Perla [Thu, 10 Nov 2011 19:18:00 +0000 (19:18 +0000)]
be2net: stop issuing FW cmds if any cmd times out

A FW cmd timeout (with a sufficiently large timeout value in the
order of tens of seconds) indicates an unresponsive FW. In this state
issuing further cmds and waiting for a completion will only stall the process.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: don't log more than one error on detecting EEH/UE errors
Sathya Perla [Thu, 10 Nov 2011 19:17:59 +0000 (19:17 +0000)]
be2net: don't log more than one error on detecting EEH/UE errors

Currently we're spamming error messages each time a FW cmd call is made
while in EEH/UE error state. One log msg on error detection is enough.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: stop checking the UE registers after an EEH error
Sathya Perla [Thu, 10 Nov 2011 19:17:58 +0000 (19:17 +0000)]
be2net: stop checking the UE registers after an EEH error

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: init (vf)_if_handle/vf_pmac_id to handle failure scenarios
Sathya Perla [Thu, 10 Nov 2011 19:17:57 +0000 (19:17 +0000)]
be2net: init (vf)_if_handle/vf_pmac_id to handle failure scenarios

Initialize if_handle, vf_if_handle and vf_pmac_id with "-1" so that in
failure cases when be_clear() is called, we can skip over
if_destroy/pmac_del cmds if they have not been created.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv4: PKTINFO doesnt need dst reference
Eric Dumazet [Wed, 9 Nov 2011 07:24:35 +0000 (07:24 +0000)]
ipv4: PKTINFO doesnt need dst reference

Le lundi 07 novembre 2011 à 15:33 +0100, Eric Dumazet a écrit :

> At least, in recent kernels we dont change dst->refcnt in forwarding
> patch (usinf NOREF skb->dst)
>
> One particular point is the atomic_inc(dst->refcnt) we have to perform
> when queuing an UDP packet if socket asked PKTINFO stuff (for example a
> typical DNS server has to setup this option)
>
> I have one patch somewhere that stores the information in skb->cb[] and
> avoid the atomic_{inc|dec}(dst->refcnt).
>

OK I found it, I did some extra tests and believe its ready.

[PATCH net-next] ipv4: IP_PKTINFO doesnt need dst reference

When a socket uses IP_PKTINFO notifications, we currently force a dst
reference for each received skb. Reader has to access dst to get needed
information (rt_iif & rt_spec_dst) and must release dst reference.

We also forced a dst reference if skb was put in socket backlog, even
without IP_PKTINFO handling. This happens under stress/load.

We can instead store the needed information in skb->cb[], so that only
softirq handler really access dst, improving cache hit ratios.

This removes two atomic operations per packet, and false sharing as
well.

On a benchmark using a mono threaded receiver (doing only recvmsg()
calls), I can reach 720.000 pps instead of 570.000 pps.

IP_PKTINFO is typically used by DNS servers, and any multihomed aware
UDP application.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv4: reduce percpu needs for icmpmsg mibs
Eric Dumazet [Tue, 8 Nov 2011 13:04:43 +0000 (13:04 +0000)]
ipv4: reduce percpu needs for icmpmsg mibs

Reading /proc/net/snmp on a machine with a lot of cpus is very expensive
(can be ~88000 us).

This is because ICMPMSG MIB uses 4096 bytes per cpu, and folding values
for all possible cpus can read 16 Mbytes of memory.

ICMP messages are not considered as fast path on a typical server, and
eventually few cpus handle them anyway. We can afford an atomic
operation instead of using percpu data.

This saves 4096 bytes per cpu and per network namespace.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: rename sk_clone to sk_clone_lock
Eric Dumazet [Tue, 8 Nov 2011 22:07:07 +0000 (17:07 -0500)]
net: rename sk_clone to sk_clone_lock

Make clear that sk_clone() and inet_csk_clone() return a locked socket.

Add _lock() prefix and kerneldoc.

Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosch_choke: use skb_header_pointer()
Eric Dumazet [Tue, 8 Nov 2011 10:45:04 +0000 (10:45 +0000)]
sch_choke: use skb_header_pointer()

Remove the assumption that skb_get_rxhash() makes IP header and ports
linear, and use skb_header_pointer() instead in choke_match_flow()

This permits __skb_get_rxhash() to use skb_header_pointer() eventually.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoll_temac: Add support for phy_mii_ioctl
Ricardo Ribalda [Mon, 7 Nov 2011 23:47:45 +0000 (23:47 +0000)]
ll_temac: Add support for phy_mii_ioctl

This patch enables the ioctl support for the driver. So userspace
programs like mii-tool can work.

Resend in merge window

Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: make ipv6 PKTINFO honour freebind
Maciej Żenczykowski [Mon, 7 Nov 2011 14:57:22 +0000 (14:57 +0000)]
net: make ipv6 PKTINFO honour freebind

This just makes it possible to spoof source IPv6 address on a socket
without having to create and bind a new socket for every source IP
we wish to spoof.

Signed-off-by: Maciej Żenczykowski <maze@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: make ipv6 bind honour freebind
Maciej Żenczykowski [Mon, 7 Nov 2011 14:57:21 +0000 (14:57 +0000)]
net: make ipv6 bind honour freebind

This makes native ipv6 bind follow the precedent set by:
  - native ipv4 bind behaviour
  - dual stack ipv4-mapped ipv6 bind behaviour.

This does allow an unpriviledged process to spoof its source IPv6
address, just like it currently can spoof its source IPv4 address
(for example when using UDP).

Signed-off-by: Maciej Żenczykowski <maze@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosweep the floors and convert some .get_drvinfo routines to strlcpy
Rick Jones [Mon, 7 Nov 2011 13:29:27 +0000 (13:29 +0000)]
sweep the floors and convert some .get_drvinfo routines to strlcpy

Per the mention made by Ben Hutchings that strlcpy is now the preferred
string copy routine for a .get_drvinfo routine, do a bit of floor
sweeping and convert some of the as-yet unconverted ethernet drivers to
it.

Signed-off-by: Rick Jones <rick.jones2@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosctp: fasthandoff with ASCONF at server-node
Michio Honda [Fri, 17 Jun 2011 02:22:35 +0000 (11:22 +0900)]
sctp: fasthandoff with ASCONF at server-node

Retransmit chunks to newly confirmed destination when ASCONF and
HEARTBEAT negotiation has success with a single-homed peer.

Signed-off-by: Michio Honda <micchie@sfc.wide.ad.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosctp: fasthandoff with ASCONF at mobile-node
Michio Honda [Fri, 17 Jun 2011 02:03:23 +0000 (11:03 +0900)]
sctp: fasthandoff with ASCONF at mobile-node

Fast retransmission after changing the last address
with ASCONF negotiation

Signed-off-by: Michio Honda <micchie@sfc.wide.ad.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: better pcpu data alignment
Eric Dumazet [Fri, 4 Nov 2011 23:19:28 +0000 (23:19 +0000)]
net: better pcpu data alignment

Tunnels can force an alignment of their percpu data to reduce number of
cache lines used in fast path, or read in .ndo_get_stats()

percpu_alloc() is a very fine grained allocator, so any small hole will
be used anyway.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv4: Fix inetpeer expire time information
Steffen Klassert [Tue, 11 Oct 2011 01:12:02 +0000 (01:12 +0000)]
ipv4: Fix inetpeer expire time information

As we update the learned pmtu informations on demand, we might
report a nagative expiration time value to userspace if the
pmtu informations are already expired and we have not send a
packet to that inetpeer after expiration. With this patch we
send a expire time of null to userspace after expiration
until the next packet is send to that inetpeer.

Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: min_pmtu default is 552
Eric Dumazet [Tue, 8 Nov 2011 19:21:44 +0000 (14:21 -0500)]
net: min_pmtu default is 552

Small fix in Documentation, since min_pmtu is 512 + 20 + 20 = 552

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotcp: Fix comments for Nagle algorithm
Feng King [Sat, 5 Nov 2011 04:23:23 +0000 (04:23 +0000)]
tcp: Fix comments for Nagle algorithm

TCP_NODELAY is weaker than TCP_CORK, when TCP_CORK was set, small
segments will always pass Nagle test regardless of TCP_NODELAY option.

Signed-off-by: Feng King <kinwin2008@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosunhme: Allow usage on SBI based SBus systems
oftedal [Mon, 7 Nov 2011 11:47:53 +0000 (11:47 +0000)]
sunhme: Allow usage on SBI based SBus systems

To prevent the SBus driver for Sun Happy Meal cards from being loaded for
PCI cards utilizing the same chipset, a filter was added to the probe
function in commit 0b492fce3d72d982a7981905f85484a1e1ba7fde.

The filter was implemented by checking the name of the parent node in
the OF tree. This patch extends this filter, so that the driver will
load on SBus systems that are based upon SBI SBus Bridges.

Signed-off-by: Kjetil Oftedal <oftedal@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agol2tp: fix l2tp_udp_recv_core()
Eric Dumazet [Tue, 8 Nov 2011 18:59:44 +0000 (13:59 -0500)]
l2tp: fix l2tp_udp_recv_core()

pskb_may_pull() can change skb->data, so we have to load ptr/optr at the
right place.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoFix incorrect usage of NET_IP_ALIGN
Nico Erfurth [Tue, 8 Nov 2011 07:30:40 +0000 (07:30 +0000)]
Fix incorrect usage of NET_IP_ALIGN

The driver used NET_IP_ALIGN to remove some additional padding inside of
the rx_fixup function. On many architectures NET_IP_ALIGN defaults to 2
which removed the correct amount of bytes.

On MCORE2-machines commit ea812ca1b06113597adcd8e70c0f84a413d97544
introduces a change which sets NET_IP_ALIGN to 0 by default. Which
triggered the bug on these machines.

This fix introduces a new RXW_PADDING define and uses this instead of
NET_IP_ALIGN. The name was taken from the original SMSC7500 driver which
is provided by SMSC.

Signed-off-by: Nico Erfurth <ne@erfurth.eu>
Tested-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv6: drop packets when source address is multicast
Brian Haley [Tue, 8 Nov 2011 04:41:42 +0000 (04:41 +0000)]
ipv6: drop packets when source address is multicast

RFC 4291 Section 2.7 says Multicast addresses must not be used as source
addresses in IPv6 packets - drop them on input so we don't process the
packet further.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Reported-and-Tested-by: Kumar Sanghvi <divinekumar@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agor8169: increase the delay parameter of pm_schedule_suspend
hayeswang [Mon, 7 Nov 2011 20:44:37 +0000 (20:44 +0000)]
r8169: increase the delay parameter of pm_schedule_suspend

The link down would occur when reseting PHY. And it would take about 2 ~ 5 seconds
from link down to link up. If the delay of pm_schedule_suspend is not long enough,
the device would enter runtime_suspend before link up. After link up, the device
would wake up and reset PHY again. Then, you would find the driver keep in a loop
of runtime_suspend and rumtime_resume.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoLinux 3.2-rc1
Linus Torvalds [Tue, 8 Nov 2011 00:16:02 +0000 (16:16 -0800)]
Linux 3.2-rc1

.. with new name.  Because nothing says "really solid kernel release"
like naming it after an extinct animal that just happened to be in the
news lately.

13 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux...
Linus Torvalds [Tue, 8 Nov 2011 00:14:26 +0000 (16:14 -0800)]
Merge branch 'fixes' of git://git./linux/kernel/git/tmlind/linux-omap

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap: (31 commits)
  ARM: OMAP: Fix export.h or module.h includes
  ARM: OMAP: omap_device: Include linux/export.h
  ARM: OMAP2: Fix H4 matrix keyboard warning
  ARM: OMAP1: Remove unused omap-alsa.h
  ARM: OMAP1: Fix warnings about enabling 32 KiHz timer
  ARM: OMAP2+: timer: Remove omap_device_pm_latency
  ARM: OMAP2+: clock data: Remove redundant timer clkdev
  ARM: OMAP: Devkit8000: Remove double omap_mux_init_gpio
  ARM: OMAP: usb: musb: OMAP: Delete unused function
  MAINTAINERS: Update linux-omap git repository
  ARM: OMAP: change get_context_loss_count ret value to int
  ARM: OMAP4: hsmmc: configure SDMMC1_DR0 properly
  ARM: OMAP4: hsmmc: Fix Pbias configuration on regulator OFF
  ARM: OMAP3: hwmod: fix variant registration and remove SmartReflex from common list
  ARM: OMAP: I2C: Fix omap_register_i2c_bus() return value on success
  ARM: OMAP: dmtimer: Include linux/module.h
  ARM: OMAP2+: l3-noc: Include linux/module.h
  ARM: OMAP2+: devices: Fixes for McPDM
  ARM: OMAP: Fix errors and warnings when building for one board
  ARM: OMAP3: PM: restrict erratum i443 handling to OMAP3430 only
  ...

13 years agoVFS: we need to set LOOKUP_JUMPED on mountpoint crossing
Al Viro [Mon, 7 Nov 2011 21:21:26 +0000 (21:21 +0000)]
VFS: we need to set LOOKUP_JUMPED on mountpoint crossing

Mountpoint crossing is similar to following procfs symlinks - we do
not get ->d_revalidate() called for dentry we have arrived at, with
unpleasant consequences for NFS4.

Simple way to reproduce the problem in mainline:

    cat >/tmp/a.c <<'EOF'
    #include <unistd.h>
    #include <fcntl.h>
    #include <stdio.h>
    main()
    {
            struct flock fl = {.l_type = F_RDLCK, .l_whence = SEEK_SET, .l_len = 1};
            if (fcntl(0, F_SETLK, &fl))
                    perror("setlk");
    }
    EOF
    cc /tmp/a.c -o /tmp/test

then on nfs4:

    mount --bind file1 file2
    /tmp/test < file1 # ok
    /tmp/test < file2 # spews "setlk: No locks available"...

What happens is the missing call of ->d_revalidate() after mountpoint
crossing and that's where NFS4 would issue OPEN request to server.

The fix is simple - treat mountpoint crossing the same way we deal with
following procfs-style symlinks.  I.e.  set LOOKUP_JUMPED...

Cc: stable@kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 7 Nov 2011 20:38:11 +0000 (12:38 -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 top: Fix live annotation in the --stdio interface
  perf top tui: Don't recalc column widths considering just the first page
  perf report: Add progress bar when processing time ordered events
  perf hists browser: Warn about lost events
  perf tools: Fix a typo of command name as trace-cmd
  perf hists: Fix recalculation of total_period when sorting entries
  perf header: Fix build on old systems
  perf ui browser: Handle K_RESIZE in dialog windows
  perf ui browser: No need to switch char sets that often
  perf hists browser: Use K_TIMER
  perf ui: Rename ui__warning_paranoid to ui__error_paranoid
  perf ui: Reimplement the popup windows using libslang
  perf ui: Reimplement ui__popup_menu using ui__browser
  perf ui: Reimplement ui_helpline using libslang
  perf ui: Improve handling sigwinch a bit
  perf ui progress: Reimplement using slang
  perf evlist: Fix grouping of multiple events

13 years agoMerge branch 'fixes-modulesplit' into fixes
Tony Lindgren [Mon, 7 Nov 2011 20:27:23 +0000 (12:27 -0800)]
Merge branch 'fixes-modulesplit' into fixes

13 years agoARM: OMAP: Fix export.h or module.h includes
Tony Lindgren [Mon, 7 Nov 2011 20:27:10 +0000 (12:27 -0800)]
ARM: OMAP: Fix export.h or module.h includes

Commit 32aaeffbd4a7457bf2f7448b33b5946ff2a960eb (Merge branch
'modsplit-Oct31_2011'...) caused some build errors. Fix these
and make sure we always have export.h or module.h included
for MODULE_ and EXPORT_SYMBOL users:

$ grep -rl ^MODULE_ arch/arm/*omap*/*.c | xargs \
  grep -L linux/module.h
  arch/arm/mach-omap2/dsp.c
  arch/arm/mach-omap2/mailbox.c
  arch/arm/mach-omap2/omap-iommu.c
  arch/arm/mach-omap2/smartreflex.c

Also check we either have export.h or module.h included
for the files exporting symbols:

$ grep -rl EXPORT_SYMBOL arch/arm/*omap*/*.c | xargs \
  grep -L linux/export.h | xargs grep -L linux/module.h

Cc: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Tony Lindgren <tony@atomide.com>
13 years agoARM: OMAP: omap_device: Include linux/export.h
Axel Lin [Mon, 7 Nov 2011 20:27:10 +0000 (12:27 -0800)]
ARM: OMAP: omap_device: Include linux/export.h

Include linux/export.h to fix below build warning:

  CC      arch/arm/plat-omap/omap_device.o
arch/arm/plat-omap/omap_device.c:1055: warning: data definition has no type or storage class
arch/arm/plat-omap/omap_device.c:1055: warning: type defaults to 'int' in declaration of 'EXPORT_SYMBOL'
arch/arm/plat-omap/omap_device.c:1055: warning: parameter names (without types) in function declaration

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Mon, 7 Nov 2011 18:55:33 +0000 (10:55 -0800)]
Merge git://git./linux/kernel/git/davem/net

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (47 commits)
  forcedeth: fix a few sparse warnings (variable shadowing)
  forcedeth: Improve stats counters
  forcedeth: remove unneeded stats updates
  forcedeth: Acknowledge only interrupts that are being processed
  forcedeth: fix race when unloading module
  MAINTAINERS/rds: update maintainer
  wanrouter: Remove kernel_lock annotations
  usbnet: fix oops in usbnet_start_xmit
  ixgbe: Fix compile for kernel without CONFIG_PCI_IOV defined
  etherh: Add MAINTAINERS entry for etherh
  bonding: comparing a u8 with -1 is always false
  sky2: fix regression on Yukon Optima
  netlink: clarify attribute length check documentation
  netlink: validate NLA_MSECS length
  i825xx:xscale:8390:freescale: Fix Kconfig dependancies
  macvlan: receive multicast with local address
  tg3: Update version to 3.121
  tg3: Eliminate timer race with reset_task
  tg3: Schedule at most one tg3_reset_task run
  tg3: Obtain PCI function number from device
  ...