openwrt/staging/blogic.git
15 years agoe1000: Use the instance of net_device_stats from net_device.
Ajit Khaparde [Wed, 7 Oct 2009 02:42:23 +0000 (02:42 +0000)]
e1000: Use the instance of net_device_stats from net_device.

Since net_device has an instance of net_device_stats,
we can remove the instance of this from the adapter structure.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: Use the instance of net_device_stats from net_device.
Ajit Khaparde [Wed, 7 Oct 2009 02:42:56 +0000 (02:42 +0000)]
igb: Use the instance of net_device_stats from net_device.

Since net_device has an instance of net_device_stats,
we can remove the instance of this from the adapter structure.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoixgbe: Use the instance of net_device_stats from net_device.
Ajit Khaparde [Wed, 7 Oct 2009 02:43:49 +0000 (02:43 +0000)]
ixgbe: Use the instance of net_device_stats from net_device.

Since net_device has an instance of net_device_stats,
we can remove the instance of this from the private adapter structure.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Acked-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agomyri10ge: Use the instance of net_device_stats from net_device.
Ajit Khaparde [Wed, 7 Oct 2009 02:45:02 +0000 (02:45 +0000)]
myri10ge: Use the instance of net_device_stats from net_device.

Since net_device has an instance of net_device_stats,
we can remove the instance of this from the private myri10ge_priv structure.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonetxen: Use the instance of net_device_stats from net_device.
Ajit Khaparde [Wed, 7 Oct 2009 02:45:34 +0000 (02:45 +0000)]
netxen: Use the instance of net_device_stats from net_device.

Since net_device has an instance of net_device_stats,
we can remove the instance of this from the private adapter structure.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoe1000e: Use the instance of net_device_stats from net_device.
Ajit Khaparde [Wed, 7 Oct 2009 02:44:26 +0000 (02:44 +0000)]
e1000e: Use the instance of net_device_stats from net_device.

Since net_device has an instance of net_device_stats,
we can remove the instance of this from the private adapter structure.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobridge: Allow enable/disable UFO on bridge device via ethtool
Sridhar Samudrala [Wed, 7 Oct 2009 12:41:17 +0000 (12:41 +0000)]
bridge: Allow enable/disable UFO on bridge device via ethtool

Allow enable/disable UFO on bridge device via ethtool

Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Make UFO on master device independent of attached devices
Sridhar Samudrala [Wed, 7 Oct 2009 12:24:25 +0000 (12:24 +0000)]
net: Make UFO on master device independent of attached devices

Now that software UFO is supported, UFO can be enabled on master
devices like bridge, bond even though the attached device doesn't
support this feature in hardware.

This allows UFO to be used between KVM host and guest even when a
physical interface attached to the bridge doesn't support UFO.

Signed-off-by: Sridhar Samudrala <sri@us.ibm.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoudp: dynamically size hash tables at boot time
Eric Dumazet [Wed, 7 Oct 2009 00:37:59 +0000 (00:37 +0000)]
udp: dynamically size hash tables at boot time

UDP_HTABLE_SIZE was initialy defined to 128, which is a bit small for
several setups.

4000 active UDP sockets -> 32 sockets per chain in average. An
incoming frame has to lookup all sockets to find best match, so long
chains hurt latency.

Instead of a fixed size hash table that cant be perfect for every
needs, let UDP stack choose its table size at boot time like tcp/ip
route, using alloc_large_system_hash() helper

Add an optional boot parameter, uhash_entries=x so that an admin can
force a size between 256 and 65536 if needed, like thash_entries and
rhash_entries.

dmesg logs two new lines :
[    0.647039] UDP hash table entries: 512 (order: 0, 4096 bytes)
[    0.647099] UDP Lite hash table entries: 512 (order: 0, 4096 bytes)

Maximal size on 64bit arches would be 65536 slots, ie 1 MBytes for non
debugging spinlocks.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoIPv6: Fix 6RD typo
Alexandre Cassen [Wed, 7 Oct 2009 21:50:30 +0000 (14:50 -0700)]
IPv6: Fix 6RD typo

Following fix a small typo.

Signed-off-by: Alexandre Cassen <acassen@freebox.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv4: Define cipso_v4_delopt static
Hagen Paul Pfeifer [Wed, 7 Oct 2009 21:45:58 +0000 (14:45 -0700)]
ipv4: Define cipso_v4_delopt static

There is no reason that cipso_v4_delopt() is not
defined as a static function.

Signed-off-by: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoeconet: Fix redeclaration of symbol len
Hagen Paul Pfeifer [Wed, 7 Oct 2009 21:43:04 +0000 (14:43 -0700)]
econet: Fix redeclaration of symbol len

Function argument len was redeclarated within the
function. This patch fix the redeclaration of symbol 'len'.

Signed-off-by: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobonding: remove useless assignment
Nicolas de Pesloüan [Wed, 7 Oct 2009 21:11:00 +0000 (14:11 -0700)]
bonding: remove useless assignment

The variable old_active is first set to bond->curr_active_slave.
Then, it is unconditionally set to new_active, without being used in between.

The first assignment, having no side effect, is useless.

Signed-off-by: Nicolas de Pesloüan <nicolas.2p.debian@free.fr>
Reviewed-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobonding: fix a parameter name in error message
Nicolas de Pesloüan [Wed, 7 Oct 2009 21:10:36 +0000 (14:10 -0700)]
bonding: fix a parameter name in error message

When parsing module parameters, bond_check_params() erroneously use
'xor_mode' as the name of a module parameter in an error message.

The right name for this parameter is 'xmit_hash_policy'.

Signed-off-by: Nicolas de Pesloüan <nicolas.2p.debian@free.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoIPv6: use ipv6_addr_set_v4mapped()
Brian Haley [Wed, 7 Oct 2009 20:58:25 +0000 (13:58 -0700)]
IPv6: use ipv6_addr_set_v4mapped()

Might as well use the ipv6_addr_set_v4mapped() inline we created last
year.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoIPv6: use ipv6_addr_copy() in ip6_route_redirect()
Brian Haley [Wed, 7 Oct 2009 20:58:01 +0000 (13:58 -0700)]
IPv6: use ipv6_addr_copy() in ip6_route_redirect()

Change ip6_route_redirect() to use ipv6_addr_copy().

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoIPv6: Fix 6RD build error
Brian Haley [Wed, 7 Oct 2009 20:57:10 +0000 (13:57 -0700)]
IPv6: Fix 6RD build error

Fix build error introduced in commit fa857afcf - ipv6 sit: 6rd
(IPv6 Rapid Deployment) Support.  Struct in6_addr is the issue.
I'm only seeing this on x86_64 systems, not on 32-bit with same
IPv6 config options, so it could be there's a missing forward
declaration somewhere, but including the correct header file
fixes the problem too.

  CC [M]  net/ipv6/ip6_tunnel.o
In file included from net/ipv6/ip6_tunnel.c:31:
include/linux/if_tunnel.h:59: error: field ‘prefix’ has incomplete type
make[2]: *** [net/ipv6/ip6_tunnel.o] Error 1
make[1]: *** [net/ipv6] Error 2

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Add sk_mark route lookup support for IPv4 listening sockets
Atis Elsts [Wed, 7 Oct 2009 20:55:57 +0000 (13:55 -0700)]
net: Add sk_mark route lookup support for IPv4 listening sockets

Add support for route lookup using sk_mark on IPv4 listening sockets.

Signed-off-by: Atis Elsts <atis@mikrotik.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agodccp ccid-3: Remove CCID naming redundancy 2/2
Gerrit Renker [Mon, 5 Oct 2009 00:53:13 +0000 (00:53 +0000)]
dccp ccid-3: Remove CCID naming redundancy 2/2

This continues the previous patch, by applying the same change to CCID-3.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agodccp ccid-2: Remove CCID naming redundancy 1/2
Gerrit Renker [Mon, 5 Oct 2009 00:53:12 +0000 (00:53 +0000)]
dccp ccid-2: Remove CCID naming redundancy 1/2

This removes a redundancy in the CCID half-connection (hc) naming scheme:
 * instead of 'hctx->tx_...', write 'hc->tx_...';
 * instead of 'hcrx->rx_...', write 'hc->rx_...';

which works because the 'type' of the half-connection is encoded in the
'rx_' / 'tx_' prefixes.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agodccp ccid-3: Overhaul CCID naming convention 2/2
Gerrit Renker [Mon, 5 Oct 2009 00:53:11 +0000 (00:53 +0000)]
dccp ccid-3: Overhaul CCID naming convention 2/2

This implements the new naming scheme also for CCID-3.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agodccp ccid-2: Overhaul CCID naming convention 1/2
Gerrit Renker [Mon, 5 Oct 2009 00:53:10 +0000 (00:53 +0000)]
dccp ccid-2: Overhaul CCID naming convention 1/2

This patch starts a less problematic naming convention for CCID structs.

The old naming convention used 'hc{tx,rx}->ccid?hc{tx,rx}->...' as
recurring prefixes, which made the code
 * hard to write (not easy to fit into 80 characters);
 * hard to read  (most of the space is occupied by prefixes).

The new naming scheme:
 * struct entries for the TX socket are prefixed by 'tx_';
 * and those for the RX socket are prefixed by 'rx_'.

The identifiers then remain distinguishable when grep-ing through the tree:
 (a) RX/TX sockets are distinguished by the naming scheme,
 (b) individual CCIDs are distinguished by filename (ccid{2,3,4}.{c,h}).

This first patch implements the scheme for CCID-2.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobe2net: Get rid of net_device_stats from adapter.
Ajit Khaparde [Wed, 7 Oct 2009 10:11:20 +0000 (03:11 -0700)]
be2net: Get rid of net_device_stats from adapter.

adapter doesnot need to maintain a copy of net_device_stats.
Use the one already available in net_device. This patch takes care of the same.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agodcb: data center bridging ops should be r/o
Stephen Hemminger [Mon, 5 Oct 2009 06:01:03 +0000 (06:01 +0000)]
dcb: data center bridging ops should be r/o

The data center bridging ops structure can be const

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: mark net_proto_ops as const
Stephen Hemminger [Mon, 5 Oct 2009 05:58:39 +0000 (05:58 +0000)]
net: mark net_proto_ops as const

All usages of structure net_proto_ops should be declared const.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agomake TLLAO option for NA packets configurable
Octavian Purdila [Fri, 2 Oct 2009 11:39:15 +0000 (11:39 +0000)]
make TLLAO option for NA packets configurable

On Friday 02 October 2009 20:53:51 you wrote:

> This is good although I would have shortened the name.

Ah, I knew I forgot something :) Here is v4.

tavi

>From 24d96d825b9fa832b22878cc6c990d5711968734 Mon Sep 17 00:00:00 2001
From: Octavian Purdila <opurdila@ixiacom.com>
Date: Fri, 2 Oct 2009 00:51:15 +0300
Subject: [PATCH] ipv6: new sysctl for sending TLLAO with unicast NAs

Neighbor advertisements responding to unicast neighbor solicitations
did not include the target link-layer address option. This patch adds
a new sysctl option (disabled by default) which controls whether this
option should be sent even with unicast NAs.

The need for this arose because certain routers expect the TLLAO in
some situations even as a response to unicast NS packets.

Moreover, RFC 2461 recommends sending this to avoid a race condition
(section 4.4, Target link-layer address)

Signed-off-by: Cosmin Ratiu <cratiu@ixiacom.com>
Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoUse sk_mark for IPv6 routing lookups
Brian Haley [Mon, 5 Oct 2009 08:24:16 +0000 (08:24 +0000)]
Use sk_mark for IPv6 routing lookups

Atis Elsts wrote:
> Not sure if there is need to fill the mark from skb in tunnel xmit functions. In any case, it's not done for GRE or IPIP tunnels at the moment.

Ok, I'll just drop that part, I'm not sure what should be done in this case.

> Also, in this patch you are doing that for SIT (v6-in-v4) tunnels only, and not doing it for v4-in-v6 or v6-in-v6 tunnels. Any reason for that?

I just sent that patch out too quickly, here's a better one with the updates.

Add support for IPv6 route lookups using sk_mark.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoethtool: Add reset operation
Ben Hutchings [Mon, 5 Oct 2009 10:59:58 +0000 (10:59 +0000)]
ethtool: Add reset operation

After updating firmware stored in flash, users may wish to reset the
relevant hardware and start the new firmware immediately.  This should
not be completely automatic as it may be disruptive.

A selective reset may also be useful for debugging or diagnostics.

This adds a separate reset operation which takes flags indicating the
components to be reset.  Drivers are allowed to reset only a subset of
those requested, and must indicate the actual subset.  This allows the
use of generic component masks and some future expansion.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agopkt_sched: gen_estimator: Dont report fake rate estimators
Eric Dumazet [Fri, 2 Oct 2009 10:32:18 +0000 (10:32 +0000)]
pkt_sched: gen_estimator: Dont report fake rate estimators

Jarek Poplawski a écrit :
>
>
> Hmm... So you made me to do some "real" work here, and guess what?:
> there is one serious checkpatch warning! ;-) Plus, this new parameter
> should be added to the function description. Otherwise:
> Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
>
> Thanks,
> Jarek P.
>
> PS: I guess full "Don't" would show we really mean it...

Okay :) Here is the last round, before the night !

Thanks again

[RFC] pkt_sched: gen_estimator: Don't report fake rate estimators

We currently send TCA_STATS_RATE_EST elements to netlink users, even if no estimator
is running.

# tc -s -d qdisc
qdisc pfifo_fast 0: dev eth0 root bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 112833764978 bytes 1495081739 pkt (dropped 0, overlimits 0 requeues 0)
 rate 0bit 0pps backlog 0b 0p requeues 0

User has no way to tell if the "rate 0bit 0pps" is a real estimation, or a fake
one (because no estimator is active)

After this patch, tc command output is :
$ tc -s -d qdisc
qdisc pfifo_fast 0: dev eth0 root bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 561075 bytes 1196 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0

We add a parameter to gnet_stats_copy_rate_est() function so that
it can use gen_estimator_active(bstats, r), as suggested by Jarek.

This parameter can be NULL if check is not necessary, (htb for
example has a mandatory rate estimator)

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet/ppp: fix comments - ppp_{sync,asynctty}_receive() may sleep
Tilman Schmidt [Thu, 1 Oct 2009 04:28:44 +0000 (04:28 +0000)]
net/ppp: fix comments - ppp_{sync,asynctty}_receive() may sleep

The receive_buf methods of the N_PPP and N_SYNC_PPP line disciplines,
ppp_asynctty_receive() and ppp_sync_receive(), call tty_unthrottle()
which may sleep. Fix the comments claiming otherwise.

Impact: documentation
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobonding: introduce primary_reselect option
Jiri Pirko [Fri, 25 Sep 2009 03:28:09 +0000 (03:28 +0000)]
bonding: introduce primary_reselect option

In some cases there is not desirable to switch back to primary interface when
it's link recovers and rather stay with currently active one. We need to avoid
packetloss as much as we can in some cases. This is solved by introducing
primary_reselect option. Note that enslaved primary slave is set as current
active no matter what.

Patch modified by Jay Vosburgh as follows: fixed bug in action
after change of option setting via sysfs, revised the documentation
update, and bumped the bonding version number.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoUse sk_mark for routing lookup in more places
Eric Dumazet [Thu, 1 Oct 2009 19:14:46 +0000 (19:14 +0000)]
Use sk_mark for routing lookup in more places

Here is a followup on this area, thanks.

[RFC] af_packet: fill skb->mark at xmit

skb->mark may be used by classifiers, so fill it in case user
set a SO_MARK option on socket.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv6 sit: 6rd (IPv6 Rapid Deployment) Support.
YOSHIFUJI Hideaki / 吉藤英明 [Tue, 22 Sep 2009 23:43:14 +0000 (23:43 +0000)]
ipv6 sit: 6rd (IPv6 Rapid Deployment) Support.

IPv6 Rapid Deployment (6rd; draft-ietf-softwire-ipv6-6rd) builds upon
mechanisms of 6to4 (RFC3056) to enable a service provider to rapidly
deploy IPv6 unicast service to IPv4 sites to which it provides
customer premise equipment.  Like 6to4, it utilizes stateless IPv6 in
IPv4 encapsulation in order to transit IPv4-only network
infrastructure.  Unlike 6to4, a 6rd service provider uses an IPv6
prefix of its own in place of the fixed 6to4 prefix.

With this option enabled, the SIT driver offers 6rd functionality by
providing additional ioctl API to configure the IPv6 Prefix for in
stead of static 2002::/16 for 6to4.

Original patch was done by Alexandre Cassen <acassen@freebox.fr>
based on old Internet-Draft.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoadd vif using local interface index instead of IP
Ilia K [Wed, 16 Sep 2009 05:53:07 +0000 (05:53 +0000)]
add vif using local interface index instead of IP

When routing daemon wants to enable forwarding of multicast traffic it
performs something like:

       struct vifctl vc = {
               .vifc_vifi  = 1,
               .vifc_flags = 0,
               .vifc_threshold = 1,
               .vifc_rate_limit = 0,
               .vifc_lcl_addr = ip, /* <--- ip address of physical
interface, e.g. eth0 */
               .vifc_rmt_addr.s_addr = htonl(INADDR_ANY),
         };
       setsockopt(fd, IPPROTO_IP, MRT_ADD_VIF, &vc, sizeof(vc));

This leads (in the kernel) to calling  vif_add() function call which
search the (physical) device using assigned IP address:
       dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr);

The current API (struct vifctl) does not allow to specify an
interface other way than using it's IP, and if there are more than a
single interface with specified IP only the first one will be found.

The attached patch (against 2.6.30.4) allows to specify an interface
by its index, instead of IP address:

       struct vifctl vc = {
               .vifc_vifi  = 1,
               .vifc_flags = VIFF_USE_IFINDEX,   /* NEW */
               .vifc_threshold = 1,
               .vifc_rate_limit = 0,
               .vifc_lcl_ifindex = if_nametoindex("eth0"),   /* NEW */
               .vifc_rmt_addr.s_addr = htonl(INADDR_ANY),
         };
       setsockopt(fd, IPPROTO_IP, MRT_ADD_VIF, &vc, sizeof(vc));

Signed-off-by: Ilia K. <mail4ilia@gmail.com>
=== modified file 'include/linux/mroute.h'
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: add Kernel CAPI interface (v3)
Tilman Schmidt [Tue, 6 Oct 2009 12:19:17 +0000 (12:19 +0000)]
gigaset: add Kernel CAPI interface (v3)

Add a Kernel CAPI interface to the Gigaset driver.

Impact: optional new functionality
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: allow building without I4L
Tilman Schmidt [Tue, 6 Oct 2009 12:19:12 +0000 (12:19 +0000)]
gigaset: allow building without I4L

Add a dummy LL interface to the Gigaset driver so that it can be
built and, in a limited way, used without the ISDN4Linux subsystem.

Impact: new configuration alternative
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: prepare for CAPI implementation
Tilman Schmidt [Tue, 6 Oct 2009 12:19:07 +0000 (12:19 +0000)]
gigaset: prepare for CAPI implementation

Reorganize the code of the Gigaset driver, moving all isdn4linux
dependencies to the source file i4l.c so that it can be replaced
by a file capi.c interfacing to Kernel CAPI instead.

Impact: refactoring, no functional change
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
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 [Wed, 7 Oct 2009 05:43:16 +0000 (22:43 -0700)]
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

15 years agogigaset: add kerneldoc comments
Tilman Schmidt [Tue, 6 Oct 2009 12:19:01 +0000 (12:19 +0000)]
gigaset: add kerneldoc comments

Add kerneldoc comments to some functions in the Gigaset driver.

Impact: documentation
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: correct debugging output selection
Tilman Schmidt [Tue, 6 Oct 2009 12:18:56 +0000 (12:18 +0000)]
gigaset: correct debugging output selection

Dump payload data consistently only when DEBUG_STREAM_DUMP debug bit
is set.

Impact: debugging aid
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: improve error recovery
Tilman Schmidt [Tue, 6 Oct 2009 12:18:51 +0000 (12:18 +0000)]
gigaset: improve error recovery

When the Gigaset base stops responding, try resetting the USB
connection to recover.

Impact: error handling improvement
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: fix device ERROR response handling
Tilman Schmidt [Tue, 6 Oct 2009 12:18:46 +0000 (12:18 +0000)]
gigaset: fix device ERROR response handling

Clear out pending command that got rejected with 'ERROR' response.
This fixes the bug where unloading the driver module would hang
with the message: "gigaset: not searching scheduled commands: busy"
after a device communication error.

Impact: error handling bugfix
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: announce if built with debugging
Tilman Schmidt [Tue, 6 Oct 2009 12:18:41 +0000 (12:18 +0000)]
gigaset: announce if built with debugging

Mention in the driver load announcement whether the driver was built
with debugging messages enabled, to facilitate support.

Impact: informational message
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: handle isoc frame errors more gracefully
Tilman Schmidt [Tue, 6 Oct 2009 12:18:36 +0000 (12:18 +0000)]
gigaset: handle isoc frame errors more gracefully

Don't drop the remainder of an URB if an isochronous frame has an error.

Impact: error handling improvement
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: linearize skb
Tilman Schmidt [Tue, 6 Oct 2009 12:18:31 +0000 (12:18 +0000)]
gigaset: linearize skb

The code of the Gigaset driver assumes that sk_buff-s coming
from the ISDN4Linux subsystem are always linear. Explicitly
calling skb_linearize() is cheap if they are, but much more
robust in case they ever aren't.

Impact: robustness improvement
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agogigaset: fix reject/hangup handling
Tilman Schmidt [Tue, 6 Oct 2009 12:18:26 +0000 (12:18 +0000)]
gigaset: fix reject/hangup handling

Signal D channel disconnect in a few cases where it was missed,
including when an incoming call is disconnected before it was
accepted.

Impact: error handling improvement
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoisdn: make capidrv module parameter "debugmode" writeable
Tilman Schmidt [Tue, 6 Oct 2009 12:18:15 +0000 (12:18 +0000)]
isdn: make capidrv module parameter "debugmode" writeable

Being able to change the debugmode module parameter of capidrv on the
fly is quite useful for debugging and doesn't do any harm.

Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoisdn: avoid races in capidrv
Tilman Schmidt [Tue, 6 Oct 2009 12:18:10 +0000 (12:18 +0000)]
isdn: avoid races in capidrv

In several places, capidrv sends a CAPI message to the ISDN
device and then updates its internal state accordingly.
If the response message from the device arrives before the
state is updated, it may be rejected or processed incorrectly.
Avoid these races by updating the state before emitting the
message.

Impact: bugfix
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Acked-by: Karsten Keil <keil@b1-systems.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoisdn: accept CAPI Informational Info values as success
Tilman Schmidt [Tue, 6 Oct 2009 12:18:05 +0000 (12:18 +0000)]
isdn: accept CAPI Informational Info values as success

Info values in the 0x00xx range are defined in the CAPI standard
as "Informational, message processed successfully". Therefore a
CONNECT_B3_CONF message with an Info value in that range should
open an NCCI just as with Info==0.

Impact: minor bugfix
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Acked-by: Karsten Keil <keil@b1-systems.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoDocumentation: expand isdn/INTERFACE.CAPI document
Tilman Schmidt [Tue, 6 Oct 2009 12:18:00 +0000 (12:18 +0000)]
Documentation: expand isdn/INTERFACE.CAPI document

- Note that send_message() may be called in interrupt context.
- Describe the storage of CAPI messages and payload data in SKBs.
- Add more details to the description of the _cmsg structure.
- Describe kernelcapi debugging output.

Impact: documentation
Signed-off-by: Tilman Schmidt <tilman@imap.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobe2net: Bug fix to properly update ethtool tx-checksumming after ethtool -K <ifname...
Ajit Khaparde [Mon, 5 Oct 2009 02:22:19 +0000 (02:22 +0000)]
be2net: Bug fix to properly update ethtool tx-checksumming after ethtool -K <ifname> tx off

This is a fix for a bug which was a result of wrong use of checksum offload flag.
The status of tx-checksumming was not changed from on to off
after a 'ethtool -K <ifname> tx off' operation.
Use the proper checksum offload flag NETIF_F_HW_CSUM instead of
NETIF_F_IP_CSUM and NETIF_F_IPV6_CSUM.
Patch is against net-2.6 tree.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobe2net: Fix a typo in be_cmds.h
Ajit Khaparde [Mon, 5 Oct 2009 02:22:05 +0000 (02:22 +0000)]
be2net: Fix a typo in be_cmds.h

MCC_STATUS_NOT_SUPPORTED should be decimal 66 not hex 66.
This patch fixes this typo. Patch against net-2.6 tree.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobe2net: Bug Fix while accounting of multicast frames during netdev stats update
Ajit Khaparde [Mon, 5 Oct 2009 02:21:51 +0000 (02:21 +0000)]
be2net: Bug Fix while accounting of multicast frames during netdev stats update

While updating the statistics to be passed via the get_stats,
tx multicast frames were being accounted instead of rx multicast frames.
This patch fixes the bug. This patch is against the net-2.6 tree.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqlge: Fix lock/mutex warnings.
Ron Mercer [Mon, 5 Oct 2009 11:46:49 +0000 (11:46 +0000)]
qlge: Fix lock/mutex warnings.

Get rid of spinlock and private mutex usage for exclusive access to the
HW semaphore register.  rtnl_lock already creates exclusive access to
this register in all driver API.
Add rtnl to firmware worker threads that also use the HW semaphore register.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqlge: Fix queueing of firmware handler in ISR.
Ron Mercer [Mon, 5 Oct 2009 11:46:48 +0000 (11:46 +0000)]
qlge: Fix queueing of firmware handler in ISR.

Check that we are not already polling firmware events before we queue the
firmware event worker, then disable firmware interrupts.
Otherwise we can queue the same event multiple times.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqlge: Fix some bit definitions for reset register.
Ron Mercer [Mon, 5 Oct 2009 11:46:47 +0000 (11:46 +0000)]
qlge: Fix some bit definitions for reset register.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: speedup sk_wake_async()
Eric Dumazet [Wed, 7 Oct 2009 00:28:29 +0000 (17:28 -0700)]
net: speedup sk_wake_async()

An incoming datagram must bring into cpu cache *lot* of cache lines,
in particular : (other parts omitted (hash chains, ip route cache...))

On 32bit arches :

offsetof(struct sock, sk_rcvbuf)       =0x30    (read)
offsetof(struct sock, sk_lock)         =0x34   (rw)

offsetof(struct sock, sk_sleep)        =0x50 (read)
offsetof(struct sock, sk_rmem_alloc)   =0x64   (rw)
offsetof(struct sock, sk_receive_queue)=0x74   (rw)

offsetof(struct sock, sk_forward_alloc)=0x98   (rw)

offsetof(struct sock, sk_callback_lock)=0xcc    (rw)
offsetof(struct sock, sk_drops)        =0xd8 (read if we add dropcount support, rw if frame dropped)
offsetof(struct sock, sk_filter)       =0xf8    (read)

offsetof(struct sock, sk_socket)       =0x138 (read)

offsetof(struct sock, sk_data_ready)   =0x15c   (read)

We can avoid sk->sk_socket and socket->fasync_list referencing on sockets
with no fasync() structures. (socket->fasync_list ptr is probably already in cache
because it shares a cache line with socket->wait, ie location pointed by sk->sk_sleep)

This avoids one cache line load per incoming packet for common cases (no fasync())

We can leave (or even move in a future patch) sk->sk_socket in a cold location

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Version update.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:59:41 +0000 (01:59 +0000)]
vxge: Version update.

- Version Update.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Allow multiple functions with INTA.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:57:29 +0000 (01:57 +0000)]
vxge: Allow multiple functions with INTA.

- Allow multiple functions with INTA.
- Removed the condition to allow only one vpath with INTA
- Ensure that the alarm bit in titan_mask_all_int register is cleared when
  driver exits.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Check if FCS stripping is disabled by the firmware.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:56:35 +0000 (01:56 +0000)]
vxge: Check if FCS stripping is disabled by the firmware.

- Added a function to check if FCS stripping is disabled by the firmware, if
  it is not disabled fail driver load.

- By default FCS stripping is disabled by the firmware. With this assumption
  driver decrements the indicated packet length by 4 bytes(FCS length).

- This patch ensures that FCS stripping is disabled during driver load time.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Removed unused functions.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:55:47 +0000 (01:55 +0000)]
vxge: Removed unused functions.

- Removed the wrr_rebalance function

- This feature is not supported by the ASIC, hence removing the related code.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Fixed crash in PAE system due to wrong typecasting.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:54:42 +0000 (01:54 +0000)]
vxge: Fixed crash in PAE system due to wrong typecasting.

- Fix a crash in PAE system due to wrong typecasting.

- On PAE system size_t is unsigned int which is 32bit. Avoid casting
  64 bit address to 32 bit

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Removed accessing non-supported registers.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:53:38 +0000 (01:53 +0000)]
vxge: Removed accessing non-supported registers.

- Removed accessing GENDMA_INT register
 - This allowed the firmware to perform a generic DMA write to host memory.
   This feature is not supported by the ASIC, this patch removes access to
   GENDMA_INT register.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Update driver_config->vpath_per_dev for each function in probe.
Sreenivasa Honnur [Mon, 5 Oct 2009 01:52:54 +0000 (01:52 +0000)]
vxge: Update driver_config->vpath_per_dev for each function in probe.

- Update driver_config->vpath_per_dev for each function in probe.

- vpath_per_device specifies number of vpaths supported for each function/device. The
  current code was updating vpath_per_device only for physical device, however this has
  to be updated for each function also in case of a MF(Multi function) device.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agovxge: Modify __vxge_hw_device_is_privilaged() to not assume function-0 as the privile...
Sreenivasa Honnur [Mon, 5 Oct 2009 01:51:38 +0000 (01:51 +0000)]
vxge: Modify __vxge_hw_device_is_privilaged() to not assume function-0 as the privileged function: Resubmit#1

- vxge driver was assuming function-0 is always the privilaged function. Now that
  restriction has been removed any function can act as a privilaged function.

- This patch modifies the __vxge_hw_device_is_privilaged routine to not assume
  function-0 as the privileged function.

- Recreated the patch by incorporating review comments from Dave Miller to
  remove double slash in path names.

Signed-off-by: Sreenivasa Honnur <sreenivasa.honnur@neterion.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoigb: add flushes between RAR writes when setting mac address
Alexander Duyck [Mon, 5 Oct 2009 06:36:01 +0000 (06:36 +0000)]
igb: add flushes between RAR writes when setting mac address

There are some switches that will do write combining when they see two
sequential regions written. In order to avoid any possible write combining
issues it is necessary to add a flush after writing each piece of a rar
register.

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 additional error handling to the phy code
Alexander Duyck [Mon, 5 Oct 2009 06:35:42 +0000 (06:35 +0000)]
igb: add additional error handling to the phy code

This update adds additional exception handling to the phy code to handle
situations where it may be called incorrectly.  In addition it adds some
bounds checking to the cable length checks to prevent an array overrun in
the event that the hardware returned a different value than expected.

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 code to retry a phy read in the event of failure on link check
Alexander Duyck [Mon, 5 Oct 2009 06:35:23 +0000 (06:35 +0000)]
igb: add code to retry a phy read in the event of failure on link check

This patch adds a retry to phy reads in the event of failure.  The original
code broke out of the loop on failure and this is a mistake as we should be
trying to do the read twice.

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 the generic copper link setup code into e1000_phy.c
Alexander Duyck [Mon, 5 Oct 2009 06:35:03 +0000 (06:35 +0000)]
igb: move the generic copper link setup code into e1000_phy.c

This patch moves the generic portion of the copper link setup into a
seperate function in e1000_phy.c.  This helps to reduce the size of
copper_link_setup_82575 and make it a bit more readable.

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 microwire support from igb
Alexander Duyck [Mon, 5 Oct 2009 06:34:44 +0000 (06:34 +0000)]
igb: remove microwire support from igb

igb doesn't have any devices that use a microwire interface for NVM.  As
such the code related to this can be removed.

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 how we handle alternate mac addresses
Alexander Duyck [Mon, 5 Oct 2009 06:34:25 +0000 (06:34 +0000)]
igb: change how we handle alternate mac addresses

This patch allows us to treat the alternate mac address as though it is the
physical address on the adapter.  This is accomplished by letting the
alt_mac_address function to only fail on an NVM error.  If no errors occur
and the alternate mac address is not present then RAR0 is read as the
default mac 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: fix a few items where weren't correctly setup for mbx timeout
Alexander Duyck [Mon, 5 Oct 2009 06:34:05 +0000 (06:34 +0000)]
igb: fix a few items where weren't correctly setup for mbx timeout

The mailbox timeout routines need to be updated as they were not correctly
handling the case of a mailbox timeout and could cause issues with long
delays when used.

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 function to handle mailbox lock
Alexander Duyck [Mon, 5 Oct 2009 06:33:46 +0000 (06:33 +0000)]
igb: add function to handle mailbox lock

Both the read and write mailbox functions need to acquire the mailbox lock.
Since that is the case we might as well combine both of the procedures into
one function so it is easier to maintain.

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 support for 82576NS SerDes adapter
Alexander Duyck [Mon, 5 Oct 2009 06:33:27 +0000 (06:33 +0000)]
igb: add support for 82576NS SerDes adapter

This patch adds the device ID necessary to support the 82576NS SerDes
adapter.

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 use of the uta to allow for promiscous mode filter
Alexander Duyck [Mon, 5 Oct 2009 06:33:08 +0000 (06:33 +0000)]
igb: make use of the uta to allow for promiscous mode filter

In order to support functions such as vlan tag stripping when SR-IOV is
enabled any given packet must match at least one filter.  However in the
case of promiscous mode being enabled on the PF the traffic routed to it
may not match any filters and is just sent to the PF by default.  In order
to make certain that this traffic is processed we can set all bits in the
UTA registers to create a pseudo promiscous mode filter that accepts all
packets.

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 combined function for setting rar and pool bits
Alexander Duyck [Mon, 5 Oct 2009 06:32:49 +0000 (06:32 +0000)]
igb: add combined function for setting rar and pool bits

This patch adds igb_rar_qsel which sets the mac address and pool bits for a
given mac address in the receive address register table.

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 locking to reads of the i2c interface
Alexander Duyck [Mon, 5 Oct 2009 06:32:27 +0000 (06:32 +0000)]
igb: add locking to reads of the i2c interface

The current implementation of sgmii support isn't correctly locking the
interfaces for reads/writes.  This change pulls the read/write
functionality out of 82575.c and moves it to phy.c.  In addition it
replaces the implementation in 82575.c with one that uses locking around
the relocated i2c interface calls.

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 the approach taken to acquiring and releasing the phy lock
Alexander Duyck [Mon, 5 Oct 2009 06:32:07 +0000 (06:32 +0000)]
igb: update the approach taken to acquiring and releasing the phy lock

The current approach is just using a ?: type mechanism to set the phy
locking bit.  This if fine for now but limits us to only 2.  Switch to a
nested if statement for future compatiblity with more than 2 phys.

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 comments for serdes config and update to handle duplex
Alexander Duyck [Mon, 5 Oct 2009 06:31:47 +0000 (06:31 +0000)]
igb: update comments for serdes config and update to handle duplex

This update corrects the driver so that it handles duplex for serdes links
correctly instead of just forcing full duplex always.

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 temp variable from stats clearing path
Alexander Duyck [Mon, 5 Oct 2009 06:31:25 +0000 (06:31 +0000)]
igb: remove unused temp variable from stats clearing path

There is a temp variable in the stats clearing path that isn't needed since
the results from the stats read can be immediately discared.  Since it
isn't needed we might as well just drop it from the function call.

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 agoconnector: Fix incompatible pointer type warning
Stephen Boyd [Tue, 6 Oct 2009 08:39:51 +0000 (01:39 -0700)]
connector: Fix incompatible pointer type warning

Commit 7069331 (connector: Provide the sender's credentials to the
callback, 2009-10-02) changed callbacks to take two arguments but missed
this one.

drivers/connector/cn_proc.c: In function ‘cn_proc_init’:
drivers/connector/cn_proc.c:263: warning: passing argument 3 of
‘cn_add_callback’ from incompatible pointer type

Signed-off-by: Stephen Boyd <bebarino@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agowext: let get_wireless_stats() sleep
Johannes Berg [Mon, 5 Oct 2009 09:22:23 +0000 (02:22 -0700)]
wext: let get_wireless_stats() sleep

A number of drivers (recently including cfg80211-based ones)
assume that all wireless handlers, including statistics, can
sleep and they often also implicitly assume that the rtnl is
held around their invocation. This is almost always true now
except when reading from sysfs:

  BUG: sleeping function called from invalid context at kernel/mutex.c:280
  in_atomic(): 1, irqs_disabled(): 0, pid: 10450, name: head
  2 locks held by head/10450:
   #0:  (&buffer->mutex){+.+.+.}, at: [<c10ceb99>] sysfs_read_file+0x24/0xf4
   #1:  (dev_base_lock){++.?..}, at: [<c12844ee>] wireless_show+0x1a/0x4c
  Pid: 10450, comm: head Not tainted 2.6.32-rc3 #1
  Call Trace:
   [<c102301c>] __might_sleep+0xf0/0xf7
   [<c1324355>] mutex_lock_nested+0x1a/0x33
   [<f8cea53b>] wdev_lock+0xd/0xf [cfg80211]
   [<f8cea58f>] cfg80211_wireless_stats+0x45/0x12d [cfg80211]
   [<c13118d6>] get_wireless_stats+0x16/0x1c
   [<c12844fe>] wireless_show+0x2a/0x4c

Fix this by using the rtnl instead of dev_base_lock.

Reported-by: Miles Lane <miles.lane@gmail.com>
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agobe2net: Fix a bug in preparation of mcc wrb which was causing flash operation to...
Ajit Khaparde [Mon, 5 Oct 2009 09:09:40 +0000 (02:09 -0700)]
be2net: Fix a bug in preparation of mcc wrb which was causing flash operation to fail

This patch fixes a bug that got introduced in commit 76998bc7.
During preparation of mcc wrb, req was being wrongly overwritten
and the flash operation was failing.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: export device speed and duplex via sysfs
Andy Gospodarek [Fri, 2 Oct 2009 09:26:12 +0000 (09:26 +0000)]
net: export device speed and duplex via sysfs

This patch exports the link-speed (in Mbps) and duplex of an interface
via sysfs.  This eliminates the need to use ethtool just to check the
link-speed.  Not requiring 'ethtool' and not relying on the SIOCETHTOOL
ioctl should be helpful in an embedded environment where space is at a
premium as well.

NOTE: This patch also intentionally allows non-root users to check the link
speed and duplex -- something not possible with ethtool.

Here's some sample output:

# cat /sys/class/net/eth0/speed
100
# cat /sys/class/net/eth0/duplex
half
# ethtool eth0
Settings for eth0:
        Supported ports: [ TP ]
        Supported link modes:   10baseT/Half 10baseT/Full
                                100baseT/Half 100baseT/Full
                                1000baseT/Half 1000baseT/Full
        Supports auto-negotiation: Yes
        Advertised link modes:  Not reported
        Advertised auto-negotiation: No
        Speed: 100Mb/s
        Duplex: Half
        Port: Twisted Pair
        PHYAD: 1
        Transceiver: internal
        Auto-negotiation: off
        Supports Wake-on: g
        Wake-on: g
        Current message level: 0x000000ff (255)
        Link detected: yes

Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agocfg80211: assign device type in netdev notifier callback
Marcel Holtmann [Fri, 2 Oct 2009 05:15:28 +0000 (05:15 +0000)]
cfg80211: assign device type in netdev notifier callback

Instead of having to modify every non-mac80211 for device type assignment,
do this inside the netdev notifier callback of cfg80211. So all drivers
that integrate with cfg80211 will export a proper device type.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: introduce NETDEV_POST_INIT notifier
Johannes Berg [Fri, 2 Oct 2009 05:15:27 +0000 (05:15 +0000)]
net: introduce NETDEV_POST_INIT notifier

For various purposes including a wireless extensions
bugfix, we need to hook into the netdev creation before
before netdev_register_kobject(). This will also ease
doing the dev type assignment that Marcel was working
on for cfg80211 drivers w/o touching them all.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agousbnet: Set device type for wlan and wwan devices
Marcel Holtmann [Fri, 2 Oct 2009 05:15:26 +0000 (05:15 +0000)]
usbnet: Set device type for wlan and wwan devices

For usbnet devices with FLAG_WLAN and FLAG_WWAN set the proper device
type so that uevent contains the correct value. This then allows an easy
identification of the actual underlying technology of the Ethernet device.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agousbnet: Use wwan%d interface name for mobile broadband devices
Marcel Holtmann [Fri, 2 Oct 2009 05:15:25 +0000 (05:15 +0000)]
usbnet: Use wwan%d interface name for mobile broadband devices

Add support for usbnet based devices like CDC-Ether to indicate that they
are actually mobile broadband devices. In that case use wwan%d as default
interface name.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agopcnet_cs: add cis of National Semicondoctor's multifunction pcmcia card
Ken Kawasaki [Mon, 5 Oct 2009 07:40:03 +0000 (00:40 -0700)]
pcnet_cs: add cis of National Semicondoctor's multifunction pcmcia card

pcnet_cs,serial_cs:

add cis of National Semicondoctor's lan&modem mulitifunction pcmcia card,
NE2K, tamarack ethernet card,
and some serial card(COMpad2, COMpad4).

Signed-off-by: Ken Kawasaki <ken_kawasaki@spring.nifty.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Support inclusion of <linux/socket.h> before <sys/socket.h>
Ben Hutchings [Mon, 5 Oct 2009 07:24:36 +0000 (00:24 -0700)]
net: Support inclusion of <linux/socket.h> before <sys/socket.h>

The following user-space program fails to compile:

    #include <linux/socket.h>
    #include <sys/socket.h>
    int main() { return 0; }

The reason is that <linux/socket.h> tests __GLIBC__ to decide whether it
should define various structures and macros that are now defined for
user-space by <sys/socket.h>, but __GLIBC__ is not defined if no libc
headers have yet been included.

It seems safe to drop support for libc 5 now.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Bastian Blank <waldi@debian.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agotunnels: Optimize tx path
Eric Dumazet [Mon, 5 Oct 2009 07:11:22 +0000 (00:11 -0700)]
tunnels: Optimize tx path

We currently dirty a cache line to update tunnel device stats
(tx_packets/tx_bytes). We better use the txq->tx_bytes/tx_packets
counters that already are present in cpu cache, in the cache
line shared with txq->_xmit_lock

This patch extends IPTUNNEL_XMIT() macro to use txq pointer
provided by the caller.

Also &tunnel->dev->stats can be replaced by &dev->stats

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoipv4: fib table algorithm performance improvement
Stephen Hemminger [Sun, 20 Sep 2009 10:35:36 +0000 (10:35 +0000)]
ipv4: fib table algorithm performance improvement

The FIB algorithim for IPV4 is set at compile time, but kernel goes through
the overhead of function call indirection at runtime. Save some
cycles by turning the indirect calls to direct calls to either
hash or trie code.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoaf_packet: add interframe drop cmsg (v6)
Neil Horman [Fri, 2 Oct 2009 06:56:41 +0000 (06:56 +0000)]
af_packet: add interframe drop cmsg (v6)

Add Ancilliary data to better represent loss information

I've had a few requests recently to provide more detail regarding frame loss
during an AF_PACKET packet capture session.  Specifically the requestors want to
see where in a packet sequence frames were lost, i.e. they want to see that 40
frames were lost between frames 302 and 303 in a packet capture file.  In order
to do this we need:

1) The kernel to export this data to user space
2) The applications to make use of it

This patch addresses item (1).  It does this by doing the following:

A) Anytime we drop a frame for which we would increment po->stats.tp_drops, we
also no increment a stats called po->stats.tp_gap.

B) Every time we successfully enqueue a frame to sk_receive_queue, we record the
value of po->stats.tp_gap in skb->mark.  skb->cb would nominally be the place to
record this, but since all the space there is used up, we're overloading
skb->mark.  Its safe to do since any enqueued packet is guaranteed to be
unshared at this point, and skb->mark isn't used for anything else in the rx
path to the application.  After we record tp_gap in the skb, we zero
po->stats.tp_gap.  This allows us to keep a counter of the number of frames lost
between any two enqueued packets

C) When the application goes to dequeue a frame from the packet socket, we look
at skb->mark for that frame.  If it is non-zero, we add a cmsg chunk to the
msghdr of level SOL_PACKET and type PACKET_GAPDATA.  Its a 32 bit integer that
represents the number of frames lost between this packet and the last previous
frame received.

Note there is a chance that if there is frame loss after a receive, and then the
socket is closed, some gap data might be lost.  This is covered by the use of
the PACKET_AUXDATA socket option, which gives total loss data.  With a bit of
math, the final gap can be determined that way.

I've tested this patch myself, and it works well.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
 include/linux/if_packet.h |    2 ++
 net/packet/af_packet.c    |   33 +++++++++++++++++++++++++++++++++
 2 files changed, 35 insertions(+)
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoTI DaVinci EMAC: Minor macro related updates
chaithrika@ti.com [Thu, 1 Oct 2009 10:25:19 +0000 (10:25 +0000)]
TI DaVinci EMAC: Minor macro related updates

Use BIT for macro definitions wherever possible, remove
unused and redundant macros.

Signed-off-by: Chaithrika U S <chaithrika@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agopktgen: Avoid dirtying skb->users when txq is full
Eric Dumazet [Wed, 30 Sep 2009 13:03:33 +0000 (13:03 +0000)]
pktgen: Avoid dirtying skb->users when txq is full

We can avoid two atomic ops on skb->users if packet is not going to be
sent to the device (because hardware txqueue is full)

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoicmp: No need to call sk_write_space()
Eric Dumazet [Thu, 24 Sep 2009 12:16:51 +0000 (12:16 +0000)]
icmp: No need to call sk_write_space()

We can make icmp messages tx completion callback a litle bit faster.

Setting SOCK_USE_WRITE_QUEUE sk flag tells sock_wfree() to
not call sk_write_space() on a socket we know no thread is posssibly
waiting for write space. (on per cpu kernel internal icmp sockets only)

This avoids the sock_def_write_space() call and
read_lock(&sk->sk_callback_lock)/read_unlock(&sk->sk_callback_lock) calls
as well.

We avoid three atomic ops.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoethtool: Remove support for obsolete string query operations
Ben Hutchings [Thu, 1 Oct 2009 11:33:03 +0000 (11:33 +0000)]
ethtool: Remove support for obsolete string query operations

The in-tree implementations have all been converted to
get_sset_count().

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agonet: Convert ethtool {get_stats, self_test}_count() ops to get_sset_count()
Ben Hutchings [Thu, 1 Oct 2009 11:58:24 +0000 (11:58 +0000)]
net: Convert ethtool {get_stats, self_test}_count() ops to get_sset_count()

These string query operations were supposed to be replaced by the
generic get_sset_count() starting in 2007.  Convert the remaining
implementations.

Also remove calls to these operations to initialise drvinfo->n_stats.
The ethtool core code already does that.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agotehuti: Convert ethtool get_stats_count() ops to get_sset_count()
Ben Hutchings [Thu, 1 Oct 2009 11:27:59 +0000 (11:27 +0000)]
tehuti: Convert ethtool get_stats_count() ops to get_sset_count()

This string query operation was supposed to be replaced by the
generic get_sset_count() starting in 2007.  Convert tehuti's
implementation.

Also remove the dummy self-test name which was not used since tehuti
does not advertise any self-tests.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
15 years agoqeth: Convert ethtool get_stats_count() ops to get_sset_count()
Ben Hutchings [Thu, 1 Oct 2009 11:24:32 +0000 (11:24 +0000)]
qeth: Convert ethtool get_stats_count() ops to get_sset_count()

This string query operation was supposed to be replaced by the
generic get_sset_count() starting in 2007.  Convert qeth's
implementation.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>