For some reason the asm versions of atomic_dec_and_test and friends were
masked by inefficient and big macros. With it I spare up to 8k in the text
size of my coldfire linux and 5k on my cpu32 version.

Patch originally submitted by Philippe De Muyter <phdm@macqel.be>
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

A number of small updates to the ColdFire serial driver:

. support the SNEHA platform
. set the default baud rate for console different on the FEITH boards
. add support for the 527x ColdFire CPU's
. schedule_timeout() should not add in current jiffies value
. optimize way that receive error characters are processed
. fix condition where receiver interrupts could be left disabled after flush
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove duplicated definition of THREAD_SIZE, it is defined in page.h.
Also modify asm constraints to explicilty let gas know that the input
arg is a constant in the current_thread_info() function.

Patch originally submitted by Philippe De Muyter <phdm@macqel.be>
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Add definitions for the SDRAM configuration registers on the 528x
ColdFire CPU's.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The 527[01] ColdFire devices and the 527[45] devices have a very similar
SDRAM register setup. But some registers are at different addresses.
Define them appropriately so common sizing code can work for all ColdFire
platforms.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Move some platform specific setup code into the ColdFire 5249 header.
Doing this for platforms that needs it means the startup code can be
identical for all ColdFire based platforms.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Change the SDRAM register names to be consistent across all ColdFire
header files. Simplied their use in common memory sizing code at
startup time.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Cache initialization code for the ColdFire CPU's. They are not
all identical. This code is used as part of the common head
start code for all ColdFire platforms.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Create a common head.S for all ColdFire architecture platforms.
Currently there is 20 (yep 20 of them), and they are 95% the
same code. This file with some header file support will replace
all of them.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Fix a stupid unbalanced lock bug in the ppc64 hugepage code.  Lead
rapidly to a crash if both CONFIG_HUGETLB_PAGE and CONFIG_PREEMPT were
enabled (even without actually using hugepages at all).
Signed-off-by: David Gibson <dwg@au1.ibm.com>
Acked-by: William Irwin <wli@holomorphy.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Include module.h to cleanup compiler warnings.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove use of obsolate MAP_NR macro.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove unused header file from include/asm-m68knommu.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Convert KTHREAD_SIZE to THREAD_SIZE.
Patch original submitted by Philippe De Muyter <phdm@macqel.be>
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Here is a patch to provide faster bitops for m68knommu, using bset/bclr/bchg
and btst instructions, that do test-and-set on m68k and coldfire processors.
We do thus avoid the need for local_irq_save/local_irq_restore.

Patch original submitted by Philippe De Muyter <phdm@macqel.be>.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Convert KTHREAD_SIZE to THREAD_SIZE.
Remove obsolate MAP_NR and VALID_PAGE macros.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Fix comment formatting in m68knommu linker script.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Export the udelay symbol so modules get access to it as well.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove duplicate M5275EVB entry.
Remove unused HIMEMKERNEL config option.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove unused keyboard.h include.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove unneccessary SUN conditionals from m68knommu elf.h.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove use of obsolete MAP_NR macro.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

into flint.arm.linux.org.uk:/usr/src/bk/linux-2.6-rmk

Patch from Nicolas Pitre

Those defines are incomplete and not using them by
instead using the explicit flags makes for clearer code.

Signed-off-by: Nicolas Pitre
Signed-off-by: Russell King

Essentially, this set of changes:
1. makes the register usage in head.S more sane
2. makes way for passing the location of the tagged list via "r2"
   (as the booting document suggests)

This brings the latency back to previous levels, and in
fact seems to drive it a bit lower.

There were a few variants on the list trying to work out what the valid
ranges to verify for write are but they sort of missed the point, if you
can load new firmware you can have fun anyway.

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

into ppc970.osdl.org:/home/torvalds/v2.6/linux

idle_task_exit() only exists when CONFIG_HOTPLUG_CPU is enabled

The patch below contains cleanups under sound/oss/ including the
following:
- make needlessly global code static
- kill cs4232.h (it contained two function prototypes: one is now static
  and the other one was already stale)
- removed the following unused global functions:
  - cs4232.c: init_cs4281
  - emu10k1/hwaccess.c: sumVolumeToAttenuation
  - emu10k1/hwaccess.c: emu10k1_set_stop_on_loop
  - sb_common.c: sb_dsp_disable_midi
  - sb_common.c: sb_dsp_disable_recording
  - wavfront.c: attach_wffx
- cs46xx.c: #ifndef CS46XX_ACPI_SUPPORT cs46xx_pm_callback
- emu10k1/hwaccess.c: #ifdef DBGEMU emu10k1_writefn0_2
- opl3sa2.c: remove the unused struct opl3sa2_dev
- removed the following unused EXPORT_SYMBOL's:
  - ac97.c: ac97_set_mixer
  - ac97.c: ac97_get_register
  - ac97.c: ac97_get_mixer_scaled
  - sb_common.c: sb_dsp_disable_midi
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

This reference to the long gone Documentation/smp.tex somehow evaded
Randy's cleanup from circa March.  This marks off the last of those stale
references.
Signed-off-by: Arthur Othieno <a.othieno@bluewin.ch>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

CT_TO_SECS() and CT_TO_USECS() are used nowhere in the tree.  This patch
removes them.
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

IN_STRING_C was an i386-specific hack by Andi Kleen to provide out-of-line
versions of some lib/string.c functions that aren't directly used on i386
(because there are inline assembly versions).

This can cause problems if gcc chooses to transparently replace a call to one
string function with a call to another string function (the example Andi gave
was a replacement of a sprintf call with a strcpy call) which generates a
function call to the replacement function.

The problems observed by Andi only occur in recent kernels without
-fno-unit-at-a-time.

With -ffreestanding it can't even cause any problems in case we'd one day drop
the -fno-unit-at-a-time since gcc mustn't assume in a freestanding environment
that all standard C library functions are available.

The effect of this patch is a small space saving in the kernel.
Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>