Reimplement handling of barrier requests.

* Flexible handling to deal with various capabilities of
  target devices.
* Retry support for falling back.
* Tagged queues which don't support ordered tag can do ordered.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJens Axboe <axboe@suse.de>

add @uptodate argument to end_that_request_last() and @error
to rq_end_io_fn().  there's no generic way to pass error code
to request completion function, making generic error handling
of non-fs request difficult (rq->errors is driver-specific and
each driver uses it differently).  this patch adds @uptodate
to end_that_request_last() and @error to rq_end_io_fn().

for fs requests, this doesn't really matter, so just using the
same uptodate argument used in the last call to
end_that_request_first() should suffice.  imho, this can also
help the generic command-carrying request jens is working on.
Signed-off-by: Ntejun heo <htejun@gmail.com>
Signed-Off-By: NJens Axboe <axboe@suse.de>

Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The original ipv6_find_hdr() finds the specified header in IPv6 packets.
This makes it possible to get transport header so that we can kill similar
loop in ip6_match_packet().
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Properly dump the helper name instead of internal kernel data.
Based on patch by Marcus Sundberg <marcus@ingate.com>.
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Driver core: add platform_device_del function

Having platform_device_del90 allows more straightforward error
handling code in drivers registering platform devices.
Signed-off-by: NDmitry Torokhov <dtor@mail.ru>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Fix build when scripts/mod/file2alias.c includes linux/input.h, which
tries to include /usr/include/linux/mod_devicetable.h:

 In file included from scripts/mod/file2alias.c:40:
 include/linux/input.h:21:35: linux/mod_devicetable.h: No such file or directory
 make[2]: *** [scripts/mod/file2alias.o] Error 1
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Here's the patch for modalias support for input classes.  It uses
comma-separated numbers, and doesn't describe all the potential keys (no
module currently cares, and that would make the strings huge).  The
changes to input.h are to move the definitions needed by file2alias
outside __KERNEL__.  I chose not to move those definitions to
mod_devicetable.h, because there are so many that it might break compile
of something else in the kernel.

The rest is fairly straightforward.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
CC: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

lib/lib.a(kobject_uevent.o)(.text+0x25f): In function `kobject_uevent':
: undefined reference to `__alloc_skb'
lib/lib.a(kobject_uevent.o)(.text+0x2a1): In function `kobject_uevent':
: undefined reference to `skb_over_panic'
lib/lib.a(kobject_uevent.o)(.text+0x31d): In function `kobject_uevent':
: undefined reference to `skb_over_panic'
lib/lib.a(kobject_uevent.o)(.text+0x356): In function `kobject_uevent':
: undefined reference to `netlink_broadcast'
lib/lib.a(kobject_uevent.o)(.init.text+0x9): In function `kobject_uevent_init':
: undefined reference to `netlink_kernel_create'
make: *** [.tmp_vmlinux1] Error 1

Netlink is unconditionally enabled if CONFIG_NET, so that's OK.

kobject_uevent.o is compiled even if !CONFIG_HOTPLUG, which is lazy.

Let's compound the sin.
Signed-off-by: NAndrew Morton <akpm@osdl.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Leave the overloaded "hotplug" word to susbsystems which are handling
real devices. The driver core does not "plug" anything, it just exports
the state to userspace and generates events.
Signed-off-by: NKay Sievers <kay.sievers@suse.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The distinction between hotplug and uevent does not make sense these
days, netlink events are the default.

udev depends entirely on netlink uevents. Only during early boot and
in initramfs, /sbin/hotplug is needed. So merge the two functions and
provide only one interface without all the options.

The netlink layer got a nice generic interface with named slots
recently, which is probably a better facility to plug events for
subsystem specific events.
Also the new poll() interface to /proc/mounts is a nicer way to
notify about changes than sending events through the core.
The uevents should only be used for driver core related requests to
userspace now.
Signed-off-by: NKay Sievers <kay.sievers@suse.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The names of these events have been confusing from the beginning
on, as they have been more like claim/release events. We needed these
events for noticing HAL if storage devices have been mounted.

Thanks to Al, we have the proper solution now and can poll()
/proc/mounts instead to get notfied about mount tree changes.
Signed-off-by: NKay Sievers <kay.sievers@suse.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

It makes zero sense to have hotplug, but not the netlink
events enabled today. Remove this option and merge the
kobject_uevent.h header into the kobject.h header file.
Signed-off-by: NKay Sievers <kay.sievers@suse.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch adds another usb-storage subdriver, which supports two fairly
old dual-XD/SmartMedia reader-writers (USB1.1 devices).

This driver was written by Daniel Drake <dsd@gentoo.org> -- he notes
that he wrote this driver without specs, however a vendor-supplied GPL
driver for the previous generation of products ("sma03") did prove to be
quite useful, as did the sddr09 driver which also has to deal with
low-level physical block layout on SmartMedia.

The original patch has been reformed by me, as it clashed with the
libusual patches.

We really need to consolidate some of this common SmartMedia code, and
get together with the MTD guys to share it with them as well.
Signed-off-by: NDaniel Drake <dsd@gentoo.org>
Signed-off-by: NMatthew Dharm <mdharm-usb@one-eyed-alien.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch (as610) adds a field to struct usb_device to store the device's
port number.  This allows us to remove several loops in the hub driver
(searching for a particular device among all the entries in the parent's
array of children).
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch (as609) changes the way we keep track of power budgeting for
USB hubs and devices, and it updates the choose_configuration routine to
take this information into account.  (This is something we should have
been doing all along.)  A new field in struct usb_device holds the amount
of bus current available from the upstream port, and the usb_hub structure
keeps track of the current available for each downstream port.

Two new rules for configuration selection are added:

	Don't select a self-powered configuration when only bus power
	is available.

	Don't select a configuration requiring more bus power than is
	available.

However the first rule is #if-ed out, because I found that the internal
hub in my HP USB keyboard claims that its only configuration is
self-powered.  The rule would prevent the configuration from being chosen,
leaving the hub & keyboard unconfigured.  Since similar descriptor errors
may turn out to be fairly common, it seemed wise not to include a rule
that would break automatic configuration unnecessarily for such devices.

The second rule may also trigger unnecessarily, although this should be
less common.  More likely it will annoy people by sometimes failing to
accept configurations that should never have been chosen in the first
place.

The patch also changes usbcore's reaction when no configuration is
suitable.  Instead of raising an error and rejecting the device, now
the core will simply leave the device unconfigured.  People can always
work around such problems by installing configurations manually through
sysfs.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch (as605) removes the private udev->serialize semaphore,
relying instead on the locking provided by the embedded struct device's
semaphore.  The changes are confined to the core, except that the
usb_trylock_device routine now uses the return convention of
down_trylock rather than down_read_trylock (they return opposite values
for no good reason).

A couple of other associated changes are included as well:

	Now that we aren't concerned about HCDs that avoid using the
	hcd glue layer, usb_disconnect no longer needs to acquire the
	usb_bus_lock -- that can be done by usb_remove_hcd where it
	belongs.

	Devices aren't locked over the same scope of code in
	usb_new_device and hub_port_connect_change as they used to be.
	This shouldn't cause any trouble.

Along with the preceding driver core patch, this needs a lot of testing.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

It is no longer needed, so let's remove it, saving a bit of memory.
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This fixes the driver that forgot to set the module owner up.  Now we
can remove the unneeded pointer from the usb driver structure.  The idea
for how to do this was from Al Viro, who did this for the PCI drivers.
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This lets drivers, like the usb-serial ones, disable the ability to add
ids from sysfs.

The usb-serial drivers are "odd" in that they are really usb-serial bus
drivers, not usb bus drivers, so the dynamic id logic will have to go
into the usb-serial bus core for those drivers to get that ability.
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Echo the usb vendor and product id to the "new_id" file in the driver's
sysfs directory, and then that driver will be able to bind to a device
with those ids if it is present.

Example:
	echo 0557 2008 > /sys/bus/usb/drivers/foo_driver/new_id
adds the hex values 0557 and 2008 to the device id table for the foo_driver.

Note, usb-serial drivers do not currently work with this capability yet.
usb-storage also might have some oddities.
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch adds a shim driver libusual, which routes devices between
usb-storage and ub according to the common table, based on unusual_devs.h.
The help and example syntax is in Kconfig.
Signed-off-by: NPete Zaitcev <zaitcev@redhat.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The check for multicast shouldn't exclude broadcast type addresses.
This reverts the incorrect change done in 2.6.13.

The broadcast address is a multicast address and should be excluded
from being a valid_ether_address for use in bridging or device address.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

From: Benjamin LaHaise <bcrl@kvack.org>

In __alloc_skb(), the use of skb_shinfo() which casts a u8 * to the 
shared info structure results in gcc being forced to do a reload of the 
pointer since it has no information on possible aliasing.  Fix this by 
using a pointer to refer to skb_shared_info.

By initializing skb_shared_info sequentially, the write combining buffers 
can reduce the number of memory transactions to a single write.  Reorder 
the initialization in __alloc_skb() to match the structure definition.  
There is also an alignment issue on 64 bit systems with skb_shared_info 
by converting nr_frags to a short everything packs up nicely.

Also, pass the slab cache pointer according to the fclone flag instead 
of using two almost identical function calls.

This raises bw_unix performance up to a peak of 707KB/s when combined 
with the spinlock patch.  It should help other networking protocols, too.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

And actually, with this, the whole pppox layer can basically
be removed and subsumed into pppoe.c, no other pppox sub-protocol
implementation exists and we've had this thing for at least 4
years.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

To help in reducing the number of include dependencies, several files were
touched as they were getting needed headers indirectly for stuff they use.

Thanks also to Alan Menegotto for pointing out that net/dccp/proto.c had
linux/dccp.h include twice.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

I noticed that some of 'struct proto_ops' used in the kernel may share
a cache line used by locks or other heavily modified data. (default
linker alignement is 32 bytes, and L1_CACHE_LINE is 64 or 128 at
least)

This patch makes sure a 'struct proto_ops' can be declared as const,
so that all cpus can share all parts of it without false sharing.

This is not mandatory : a driver can still use a read/write structure
if it needs to (and eventually a __read_mostly)

I made a global stubstitute to change all existing occurences to make
them const.

This should reduce the possibility of false sharing on SMP, and
speedup some socket system calls.
Signed-off-by: NEric Dumazet <dada1@cosmosbay.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sock_init can be done as a core_initcall instead of calling
it directly in init/main.c

Also I removed an out of date #ifdef.
Signed-off-by: NAndi Kleen <ak@suse.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Here is a new feature for netem in 2.6.16. It adds the ability to
randomly corrupt packets with netem. A version was done by
Hagen Paul Pfeifer, but I redid it to handle the cases of backwards
compatibility with netlink interface and presence of hardware checksum
offload. It is useful for testing hardware offload in devices.
Signed-off-by: NStephen Hemminger <shemminger@osdl.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As DCCP needs to be called in the same spots.

Now we have a member in inet_sock (is_icsk), set at sock creation time from
struct inet_protosw->flags (if INET_PROTOSW_ICSK is set, like for TCP and
DCCP) to see if a struct sock instance is a inet_connection_sock for places
like the ones in ip_sockglue.c (v4 and v6) where we previously were looking if
sk_type was SOCK_STREAM, that is insufficient because we now use the same code
for DCCP, that has sk_type SOCK_DCCP.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Upcoming patches will make, for instance, ip_sockglue.c need just this enum
and not all of tcp.h.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Renaming it to inet6_hash_connect, making it possible to ditch
dccp_v6_hash_connect and share the same code with TCP instead.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Renaming it to inet_hash_connect, making it possible to ditch
dccp_v4_hash_connect and share the same code with TCP instead.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

So that we can share several timewait sockets related functions and
make the timewait mini sockets infrastructure closer to the request
mini sockets one.

Next changesets will take advantage of this, moving more code out of
TCP and DCCP v4 and v6 to common infrastructure.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Out of tcp6_timewait_sock, that now is just an aggregation of
inet_timewait_sock and inet6_timewait_sock, using tw_ipv6_offset in struct
inet_timewait_sock, that is common to the IPv6 transport protocols that use
timewait sockets, like DCCP and TCP.

tw_ipv6_offset plays the struct inet_sock pinfo6 role, i.e. for the generic
code to find the IPv6 area in a timewait sock.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using sk->sk_protocol instead of IPPROTO_TCP.

Will be used by DCCPv6 in the next changesets.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When a packet is obtained from skb_recv_datagram with MSG_PEEK enabled
it is left on the socket receive queue.  This means that when we detect
a checksum error we have to be careful when trying to free the packet
as someone could have dequeued it in the time being.

Currently this delicate logic is duplicated three times between UDPv4,
UDPv6 and RAWv6.  This patch moves them into a one place and simplifies
the code somewhat.

This is based on a suggestion by Eric Dumazet.
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

And move it to struct inet_connection_sock. DCCP will use it in the
upcoming changesets.
Signed-off-by: NArnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>