[PATCH] kdump: dynamic per cpu allocation of memory for saving cpu registers

- In case of system crash, current state of cpu registers is saved in memory
  in elf note format.  So far memory for storing elf notes was being allocated
  statically for NR_CPUS.

- This patch introduces dynamic allocation of memory for storing elf notes.
  It uses alloc_percpu() interface.  This should lead to better memory usage.

- Introduced based on Andi Kleen's and Eric W. Biederman's suggestions.

- This patch also moves memory allocation for elf notes from architecture
  dependent portion to architecture independent portion.  Now crash_notes is
  architecture independent.  The whole idea is that size of memory to be
  allocated per cpu (MAX_NOTE_BYTES) can be architecture dependent and
  allocation of this memory can be architecture independent.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] kdump: i386 save ss esp bug fix

\r)

From: Vivek Goyal <vgoyal@in.ibm.com>

This patch fixes a minor bug based on Andi Kleen's suggestion.  asm's can't be
broken in this particular case, hence merging them.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Disable rio on 64-bit platforms

Do it via Kconfig rather than via #error.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Remove set_fs() in stop_machine()

\r)

From: Brian Gerst <bgerst@didntduck.org>

Call sched_setscheduler() directly instead.

Signed-off-by: Brian Gerst <bgerst@didntduck.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] dump_thread() cleanup

\r)

From: Adrian Bunk <bunk@stusta.de>

- create one common dump_thread() prototype in kernel.h

- dump_thread() is only used in fs/binfmt_aout.c and can therefore be
  removed on all architectures where CONFIG_BINFMT_AOUT is not
  available

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] ipmi: fix compile errors with PROC_FS=n

...
  CC [M]  drivers/char/ipmi/ipmi_msghandler.o
drivers/char/ipmi/ipmi_msghandler.c:3301: `proc_ipmi_root' undeclared here (not in a function)
drivers/char/ipmi/ipmi_msghandler.c:3301: initializer element is not constant
drivers/char/ipmi/ipmi_msghandler.c:3301: (near initialization for `__ksymtab_proc_ipmi_root.value')
drivers/char/ipmi/ipmi_msghandler.c:1535: warning: `ipmb_file_read_proc' defined but not used
drivers/char/ipmi/ipmi_msghandler.c:1551: warning: `version_file_read_proc' defined but not used
drivers/char/ipmi/ipmi_msghandler.c:1561: warning: `stat_file_read_proc' defined but not used
...
  CC [M]  drivers/char/ipmi/ipmi_poweroff.o
drivers/char/ipmi/ipmi_poweroff.c: In function `ipmi_poweroff_init':
drivers/char/ipmi/ipmi_poweroff.c:616: warning: implicit declaration of function `unregister_sysctl_table'
drivers/char/ipmi/ipmi_poweroff.c:616: `ipmi_table_header' undeclared (first use in this function)
drivers/char/ipmi/ipmi_poweroff.c:616: (Each undeclared identifier is reported only once
drivers/char/ipmi/ipmi_poweroff.c:616: for each function it appears in.)

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Corey Minyard <minyard@acm.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] printk levels for i386 oops code.

Especially useful when users have booted with 'quiet'.  In the regular 'oops'
path, we set the console_loglevel before we start spewing debug info, but we
can call the backtrace code from other places now too, such as the spinlock
debugging code.

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] printk levels for spinlock debug

Signed-off-by: Dave Jones <davej@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Add list_for_each_entry_safe_reverse()

Add list_for_each_entry_safe_reverse() to linux/list.h

This is needed by unmerged cachefs and be an as-yet-unreviewed
device_shutdown() fix.

Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Patrick Mochel <mochel@digitalimplant.org>
Cc: Greg KH <greg@kroah.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] i386: GPIO driver for AMD CS5535/CS5536

A simple driver for the CS5535 and CS5536 that allows a user-space program
to manipulate GPIO pins.  The CS5535/CS5536 chips are Geode processor
companion devices.

Signed-off-by: Ben Gardner <bgardner@wabtec.com>
Signed-off-by: Richard Knutsson <ricknu-0@student.ltu.se>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] fix /sys/class/net/<if>/wireless without dev->get_wireless_stats

dev->get_wireless_stats is deprecated but removing it also removes wireless
subdirectory in sysfs. This patch puts it back.

akpm: I don't know what's happening here.  This might be appropriate as a
2.6.15.x compatibility backport.  Waiting to hear from Jeff.

Signed-off-by: Andrey Borzenkov <arvidjaar@mail.ru>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jeff Garzik <jgarzik@pobox.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] nvidiafb: Fixes for new G5

Recent X "nv" driver was fixed for various issues with modern 6xxx and 7xxx
cards.  This patch ports those fixes to nvidiafb.  This makes it work fine
on the 6600 bundled with the newest G5 macs.  I've verified it still works
on the 5200FX of the iMacG5.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] "tiny-make-id16-support-optional" fixes

It seems the "make UID16 support optional" patch was checked when it
edited the -tiny tree some time ago, but it wasn't checked whether it
still matches the current situation when it was submitted for inclusion
in -mm. This patch fixes the following bugs:
- ARCH_S390X does no longer exist, nowadays this has to be expressed
  through (S390 && 64BIT)
- in five architecture specific Kconfig files the UID16 options
  weren't removed

Additionally, it changes the fragile negative dependencies of UID16 to
positive dependencies (new architectures are more likely to not require
UID16 support).

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Acked-by: Matt Mackall <mpm@selenic.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] remove asm/serial.h from synclink_gt.

Not all architectures implement asm/serial.h, and the driver doesn't appear to
need it anyway.

Cc: Paul Fulghum <paulkf@microgate.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] spufs: fix for recent "shrink dentry_struct" patch

Cc: Eric Dumazet <dada1@cosmosbay.com>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Merge /pub/scm/linux/kernel/git/gregkh/pci-2.6

Merge /pub/scm/linux/kernel/git/mingo/mutex-2.6

Merge git://git./linux/kernel/git/bunk/trivial

[PATCH] rcu: don't set ->next_pending in rcu_start_batch()

I think it is better to set ->next_pending in the caller, when
it is needed. This saves one parameter, and this coincides with
cpu_quiet() beahaviour, which sets ->completed = ->cur itself.

Signed-off-by: Oleg Nesterov <oleg@tv-sign.ru>
Acked-by: Paul E. McKenney <paulmck@us.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] mutex subsystem, semaphore to completion: drivers/block/loop.c

convert the block loop device from semaphores to completions.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, semaphore to completion: IDE ->gendev_rel_sem

The patch changes semaphores that are initialized as
locked to complete().

Source: MontaVista Software, Inc.

Modified-by: Steven Rostedt <rostedt@goodmis.org>
The following patch is from Montavista.  I modified it slightly.
Semaphores are currently being used where it makes more sense for
completions.  This patch corrects that.

Signed-off-by: Aleksey Makarov <amakarov@ru.mvista.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, semaphore to completion: CPU3WDT

change CPU3WDT semaphores to completions.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, semaphore to completion: SX8

change SX8 semaphores to completions.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, semaphore to mutex: VFS, sb->s_lock

This patch converts the superblock-lock semaphore to a mutex, affecting
lock_super()/unlock_super(). Tested on ext3 and XFS.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, semaphore to mutex: VFS, ->i_sem

This patch converts the inode semaphore to a mutex. I have tested it on
XFS and compiled as much as one can consider on an ia64. Anyway your
luck with it might be different.

Modified-by: Ingo Molnar <mingo@elte.hu>
(finished the conversion)

Signed-off-by: Jes Sorensen <jes@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, semaphore to mutex: XFS

This patch switches XFS over to use the new mutex code directly as
opposed to the previous workaround patch I posted earlier that avoided
the namespace clash by forcing it back to semaphores. This falls in the
'works for me<tm>' category.

Signed-off-by: Jes Sorensen <jes@trained-monkey.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, more debugging code

more mutex debugging: check for held locks during memory freeing,
task exit, enable sysrq printouts, etc.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

[PATCH] mutex subsystem, debugging code

mutex implementation - add debugging code.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

[PATCH] mutex subsystem, documentation

Add mutex design related documentation.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

[PATCH] mutex subsystem, core

mutex implementation, core files: just the basic subsystem, no users of it.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

[PATCH] mutex subsystem, add default include/asm-*/mutex.h files

add the per-arch mutex.h files for the remaining architectures.

We default to asm-generic/mutex-dec.h, because that performs
quite well on most arches. Arches that do not have atomic
decrement/increment instructions should switch to mutex-xchg.h
instead. Arches can also provide their own implementation for
the mutex fastpath primitives.

Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, add include/asm-arm/mutex.h

add the ARM version of mutex.h, which is optimized in assembly for
ARMv6, and uses the xchg implementation on pre-ARMv6.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, add include/asm-x86_64/mutex.h

add the x86_64 version of mutex.h, optimized in assembly.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

[PATCH] mutex subsystem, add include/asm-i386/mutex.h

add the i386 version of mutex.h, optimized in assembly.

Signed-off-by: Arjan van de Ven <arjan@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, add asm-generic/mutex-[dec|xchg|null].h implementations

Add three (generic) mutex fastpath implementations.

The mutex-xchg.h implementation is atomic_xchg() based, and should
work fine on every architecture.

The mutex-dec.h implementation is atomic_dec_return() based - this
one too should work on every architecture, but might not perform the
most optimally on architectures that have no atomic-dec/inc instructions.

The mutex-null.h implementation forces all calls into the slowpath. This
is used for mutex debugging, but it can also be used on platforms that do
not want (or need) a fastpath at all.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

[PATCH] mutex subsystem, add typecheck_fn(type, function)

add typecheck_fn(type, function) to do type-checking of function
pointers.

Modified-by: Ingo Molnar <mingo@elte.hu>
(made it typeof() based, instead of typedef based.)

Signed-off-by: Chuck Ebbert <76306.1226@compuserve.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

[PATCH] mutex subsystem, add atomic_xchg() to all arches

add atomic_xchg() to all the architectures. Needed by the new mutex code.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arjan van de Ven <arjan@infradead.org>

Small fixups to the EHCI Kconfig help text

Here's a small patch with a few tiny fixups for the EHCI Kconfig help
text. Please consider applying.

Signed-off-by: Jesper Juhl <juhl-lkml@dif.dk>
Signed-off-by: Adrian Bunk <bunk@stusta.de>

Documentation/filesystems/proc.txt: indentation fix

Signed-off-by: Adrian Bunk <bunk@stusta.de>

remove the outdated arch/i386/kernel/cpu/{,mtrr/}changelog

This patch removes two outdated changelog files.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dave Jones <davej@redhat.com>

MAINTAINERS: remove BUSLOGIC entry

Leonard N. Zubkoff passed away in August 2002.

Signed-off-by: Adrian Bunk <bunk@stusta.de>

spelling: s/trough/through/

Additionally, one comment was reformulated by Joe Perches <joe@perches.com>.

Signed-off-by: Adrian Bunk <bunk@stusta.de>

Merge /pub/scm/linux/kernel/git/davem/net-2.6

Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

spelling: s/retreive/retrieve/

Signed-off-by: Adrian Bunk <bunk@stusta.de>

s/assoicated/associated/

Signed-off-by: Adrian Bunk <bunk@stusta.de>

Merge master.kernel.org:/home/rmk/linux-2.6-arm

Merge master.kernel.org:/home/rmk/linux-2.6-serial

Merge master.kernel.org:/home/rmk/linux-2.6-mmc

spelling: s/usefull/useful/

Signed-off-by: Adrian Bunk <bunk@stusta.de>

[INET_DIAG]: Introduce sk_diag_fill

To be called from inet_diag_get_exact, also rename inet_diag_fill to
inet_csk_diag_fill, for consistency with inet_twsk_diag_fill.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[INET_DIAG]: Introduce inet_twsk_diag_dump & inet_twsk_diag_fill

To properly dump TIME_WAIT sockets and to reduce complexity a bit by
having per socket class accessor routines.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[INET_DIAG]: whitespace/simple cleanups

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[INET_DIAG]: Use inet_twsk() with TIME_WAIT sockets

The fields being accessed in inet_diag_dump are outside sock_common, the
common part of struct sock and struct inet_timewait_sock.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[MMC] Indicate that R1/R1b contains command opcode

Some controllers actually check the first byte of the response (most
don't).  This byte contains the command opcode for R1/R1b and all 1:s
for other types. The difference must be indicated to the controller
so it knows which reply to expect.

Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>

[SPARC64]: Update defconfig.

Signed-off-by: David S. Miller <davem@davemloft.net>

[SPARC64]: Fix ptrace/strace

Don't clobber register %l0 while checking TI_SYS_NOERROR value in
syscall return path.  This bug was introduced by:

db7d9a4eb700be766cc9f29241483dbb1e748832

Problem narrowed down by Luis F. Ortiz and Richard Mortimer.

I tried using %l2 as suggested by Luis and that works for me.

Looking at the code I wonder if it makes sense to simplify the code
a little bit. The following works for me but I'm not sure how to
exercise the "NOERROR" codepath.

Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: Set skb->priority in ip6_output.c

Set skb->priority = sk->sk_priority as in raw.c and IPv4.

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

[IPV4]: ip_output.c needs xfrm.h

This patch fixes a warning from my IPsec patches:

   CC      net/ipv4/ip_output.o
net/ipv4/ip_output.c: In function 'ip_finish_output':
net/ipv4/ip_output.c:208: warning: implicit declaration of function
'xfrm4_output_finish'

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

[PKT_SCHED]: Fix qdisc return code.

The mapping between TC_ACTION_SHOT and the qdisc return codes is better
suited to NET_XMIT_BYPASS so as not to confuse TCP

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Add IFB (Intermediate Functional Block) network device.

A new device to do intermidiate functional block in a system shared
manner.  To use the new functionality, you need to turn on
qos/classifier actions.

The new functionality can be grouped as:

1) qdiscs/policies that are per device as opposed to system wide.  ifb
allows for a device which can be redirected to thus providing an
impression of sharing.

2) Allows for queueing incoming traffic for shaping instead of
dropping.

Packets are redirected to this device using tc/action mirred redirect
construct. If they are sent to it by plain routing instead then they
will merely be dropped and the stats would indicate that.

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[AX25] mkiss: Drop spinlock before sleeping call.

With the previous missing-unlock fix the spinlock is dropped only
after the tty->driver->write() call which might sleep.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Change some "if (x) BUG();" to "BUG_ON(x);"

This changes some simple "if (x) BUG();" statements to "BUG_ON(x);"

Signed-off-by: Kris Katterjohn <kjak@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

[PKT_SCHED]: Prefix tc actions with act_

Clean up the net/sched directory a bit by prefix all actions with act_.

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

[PKT_SCHED]: Fix memory leak when dumping in pedit action

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

[PKT_SCHED]: Remove some obsolete policer exports

Also make sure the legacy code is only built when CONFIG_NET_CLS_ACT
is not set.

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

[PKT_SCHED]: Convert tc action functions to single skb pointers

tcf_action_exec only gets a single skb pointer and doesn't own the skb,
but passes double skb pointers (to a local variable) to the action
functions. Change to use single skb pointers everywhere.

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

[PKT_SCHED]: Use USEC_PER_SEC

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

[NET]: Convert net/{ipv4,ipv6,sched} to netdev_priv

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

[CRYPTO] cipher: Set alignmask for multi-byte loads

Many cipher implementations use 4-byte/8-byte loads/stores which require
alignment on some architectures.  This patch explicitly sets the alignment
requirements for them.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] api: Require block size to be less than PAGE_SIZE/8

The cipher code path may allocate up to two blocks of data on the stack.
Therefore we need to place limits on the maximum block size.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] sha1: Fixed off-by-64 bug in sha1_update

After a partial update, the done pointer is off to the right by 64 bytes.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] aes-i586: Remove unused variable ls_tab

It is assigned but never read.

Signed-off-by: Daniel Marjamäki <daniel.marjamaki@comhem.se>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] aes-i586: Nano-optimisation on key length check

Reduce the number of comparisons by one through the use of jb/je.
This patch also corrects the comments regarding the different key
lengths.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] cipher: Align temporary buffer in cbc_process_decrypt

Since the temporary buffer is used as an argument to cia_decrypt, it must be
aligned by cra_alignmask.  This bug was found by linux@horizon.com.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] sha1: Avoid shifting count left and right

This patch avoids shifting the count left and right needlessly for each
call to sha1_update().  It instead can be done only once at the end in
sha1_final().

Keeping the previous test example (sha1_update() successively called with
len=64), a 1.3% performance increase can be observed on i386, or 0.2% on
ARM.  The generated code is also smaller on ARM.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] sha1: Rename i/j to done/partial

This patch gives more descriptive names to the variables i and j.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] sha1: Avoid useless memcpy()

The current code unconditionally copy the first block for every call to
sha1_update().  This can be avoided if there is no pending partial block.
This is always the case on the first call to sha1_update() (if the length
is >= 64 of course.

Furthermore, temp does need to be called if sha_transform is never invoked.
Also consolidate the sha_transform calls into one to reduce code size.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] Allow AES C/ASM implementations to coexist

As the Crypto API now allows multiple implementations to be registered
for the same algorithm, we no longer have to play tricks with Kconfig
to select the right AES implementation.

This patch sets the driver name and priority for all the AES
implementations and removes the Kconfig conditions on the C implementation
for AES.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] Allow multiple implementations of the same algorithm

This is the first step on the road towards asynchronous support in
the Crypto API.  It adds support for having multiple crypto_alg objects
for the same algorithm registered in the system.

For example, each device driver would register a crypto_alg object
for each algorithm that it supports.  While at the same time the
user may load software implementations of those same algorithms.

Users of the Crypto API may then select a specific implementation
by name, or choose any implementation for a given algorithm with
the highest priority.

The priority field is a 32-bit signed integer.  In future it will be
possible to modify it from user-space.

This also provides a solution to the problem of selecting amongst
various AES implementations, that is, aes vs. aes-i586 vs. aes-padlock.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[CRYPTO] Use standard byte order macros wherever possible

A lot of crypto code needs to read/write a 32-bit/64-bit words in a
specific gender.  Many of them open code them by reading/writing one
byte at a time.  This patch converts all the applicable usages over
to use the standard byte order macros.

This is based on a previous patch by Denis Vlasenko.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[PADLOCK] Fix sparse warning about 1-bit signed bit-field

Change the bit-field in struct cword to unsigned to shut sparse up.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>

[SPARC64]: Add needed pm_power_off symbol.

Signed-off-by: David S. Miller <davem@davemloft.net>

[ARM] 3246/1: S3C24XX - retab clock list in arch/arm/mach-s3c2410/clock.c

Patch from Ben Dooks

Properly tabulate the clock table in arch/arm/mach-s3c2410/clock.c
and put the requisite commas on the end of the structs.

Fix the comment about clock enable and disable in the setup code

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

[MMC] Add DATA_MULTI flag

Some hosts need to know that a transfer will be multi-block.
Add a data flag to indicate multiple data block transfers.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

Merge /pub/scm/linux/kernel/git/mchehab/v4l-dvb

[PATCH] Fix more "if ((err = foo() < 0))" typos

Another reason to use:

ret = foo();
if (ret < 0)
goto out;

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

[PATCH] Fix sg_page_malloc() memset

sg_page_malloc should clear the data buffer, not that extent of mem_map.

This fixes Jesper's sg_page_free "Bad page states"

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

V4L/DVB (3336): Bluebird firmware hosting moved to common dvb firmware dir on linuxtv.org

- Get_dvb_firmware script updated accordingly:
- my $url = "http://www.linuxtv.org/~mkrufky/dvb-usb-bluebird-01.fw";
+ my $url = "http://www.linuxtv.org/download/dvb/firmware/dvb-usb-blue$

Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3329): cx88_subids whitespace cleanup

- Clean up a recently introduced spaces vs tabs whitespace problem.

Signed-off-by: Chris Pascoe <c.pascoe@itee.uq.edu.au>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3328): replace omitted 'TODO:'

- replace omitted 'TODO:'

Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3327): Remove DViCO specific firmware hacks from the generic code.

- Move the code that patches bluebird firmware before upload from the
generic code into the cxusb driver itself.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Michael Krufky <mkrufky@m1k.net>
Signed-off-by: Chris Pascoe <c.pascoe@itee.uq.edu.au>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3326): Fix stereo and standard reporting of msp3400 (esp. for radio)

- Add VIDIOC_LOG_STATUS to the radio device ioctl list.
- Reduce the confusion between modes and standards
- Fix stereo reporting for radio.
- Don't set i2c configuration if the chip doesn't support it.
- Fix reporting of current standard for radio.

Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3325): WSS output interface for av7110

- Implemented v4l2 api for sliced vbi data output
to pass WSS data from userspace to the av7110

Signed-off-by: Oliver Endriss <o.endriss@gmx.de>
Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3324): msp3400 audio handling bug fixes.

- Check capabilities for audio settings (volume, balance,
bass, treble, loudness, mute)
- added loudness support
- added missing VIDEO_AUDIO_BALANCE flags for v4l1 compatibility
- do not call msp_any_detect_stereo for non-autoselect chips to
retrieve the current stereo setting: that will temporarily mute
the sound. It is only needed when the stereo mode might be
changed, and for autoselect msp processors that do not periodically
need to update their stereo setting.
- do not wake up the thread if the standard did not change. Prevents
temporary audio drop-out if the standard is set to the same value.
- fix confused stereo detect code where V4L2_TUNER_SUB_STEREO and
V4L2_TUNER_MODE_STEREO values were used incorrectly.
- stereo mode reporting was broken (v4l2 value used to index a
string array expecting v4l1 mode values).
- do not set dsp register 0x30 in the 3410d thread: that register
does not exist for pre-'G' revision msp chips.

Signed-off-by: Hans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3323): Add Kconfig option for wrongly programmed devices

- A new Kconfig option makes the user able to select if the dibusb-mb driver shall claim faulty programmed USB devices which are coming with default Cypress USB IDs.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3322): Whitespace cleanup and unused code removed

- Some whitespaces cleaned up
- unused code removed

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3321): Indention cleanups

- Indention cleaned for tabsize=8

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3320): Check if PLL-description is set

- Check if in the dvb-usb-device-structure the dvb_pll_desc is set before use.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>

V4L/DVB (3319): Changed indention for define's and their values

- Indention for values of several request defines were not TabSized=8.

Signed-off-by: Patrick Boettcher <pb@linuxtv.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@brturbo.com.br>