Provide an alternate implementation of the SLUB fast paths for alloc
and free using cmpxchg_local. The cmpxchg_local fast path is selected
for arches that have CONFIG_FAST_CMPXCHG_LOCAL set. An arch should only
set CONFIG_FAST_CMPXCHG_LOCAL if the cmpxchg_local is faster than an
interrupt enable/disable sequence. This is known to be true for both
x86 platforms so set FAST_CMPXCHG_LOCAL for both arches.

Currently another requirement for the fastpath is that the kernel is
compiled without preemption. The restriction will go away with the
introduction of a new per cpu allocator and new per cpu operations.

The advantages of a cmpxchg_local based fast path are:

1. Potentially lower cycle count (30%-60% faster)

2. There is no need to disable and enable interrupts on the fast path.
   Currently interrupts have to be disabled and enabled on every
   slab operation. This is likely avoiding a significant percentage
   of interrupt off / on sequences in the kernel.

3. The disposal of freed slabs can occur with interrupts enabled.

The alternate path is realized using #ifdef's. Several attempts to do the
same with macros and inline functions resulted in a mess (in particular due
to the strange way that local_interrupt_save() handles its argument and due
to the need to define macros/functions that sometimes disable interrupts
and sometimes do something else).

[clameter: Stripped preempt bits and disabled fastpath if preempt is enabled]
Signed-off-by: NChristoph Lameter <clameter@sgi.com>
Reviewed-by: NPekka Enberg <penberg@cs.helsinki.fi>
Cc: <linux-arch@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>

We use a NULL pointer on freelists to signal that there are no more objects.
However the NULL pointers of all slabs match in contrast to the pointers to
the real objects which are in different ranges for different slab pages.

Change the end pointer to be a pointer to the first object and set bit 0.
Every slab will then have a different end pointer. This is necessary to ensure
that end markers can be matched to the source slab during cmpxchg_local.

Bring back the use of the mapping field by SLUB since we would otherwise have
to call a relatively expensive function page_address() in __slab_alloc().  Use
of the mapping field allows avoiding a call to page_address() in various other
functions as well.

There is no need to change the page_mapping() function since bit 0 is set on
the mapping as also for anonymous pages.  page_mapping(slab_page) will
therefore still return NULL although the mapping field is overloaded.
Signed-off-by: NChristoph Lameter <clameter@sgi.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>

gcc 4.2 spits out an annoying warning if one casts a const void *
pointer to a void * pointer. No warning is generated if the
conversion is done through an assignment.
Signed-off-by: NChristoph Lameter <clameter@sgi.com>

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC32]: Use regsets in arch_ptrace().
  [SPARC64]: Use regsets in arch_ptrace().
  [SPARC32]: Use regsets for ELF core dumping.
  [SPARC64]: Use regsets for ELF core dumping.
  [SPARC64]: Remove unintentional ptrace debugging messages.
  [SPARC]: Move over to arch_ptrace().
  [SPARC]: Remove PTRACE_SUN* handling.
  [SPARC]: Kill DEBUG_PTRACE code.
  [SPARC32]: Add user regset support.
  [SPARC64]: Add user regsets.
  [SPARC64]: Fix booting on non-zero cpu.

* git://git.infradead.org/mtd-2.6: (120 commits)
  [MTD] Fix mtdoops.c compilation
  [MTD] [NOR] fix startup lock when using multiple nor flash chips
  [MTD] [DOC200x] eccbuf is statically defined and always evaluate to true
  [MTD] Fix maps/physmap.c compilation with CONFIG_PM
  [MTD] onenand: Add panic_write function to the onenand driver
  [MTD] mtdoops: Use the panic_write function when present
  [MTD] Add mtd panic_write function pointer
  [MTD] [NAND] Freescale enhanced Local Bus Controller FCM NAND support.
  [MTD] physmap.c: Add support for multiple resources
  [MTD] [NAND] Fix misparenthesization introduced by commit 78b65179...
  [MTD] [NAND] Fix Blackfin NFC ECC calculating bug with page size 512 bytes
  [MTD] [NAND] Remove wrong operation in PM function of the BF54x NFC driver
  [MTD] [NAND] Remove unused variable in plat_nand_remove
  [MTD] Unlocking all Intel flash that is locked on power up.
  [MTD] [NAND] at91_nand: Make mtdparts option can override board info
  [MTD] mtdoops: Various minor cleanups
  [MTD] mtdoops: Ensure sequential write to the buffer
  [MTD] mtdoops: Perform write operations in a workqueue
  [MTD] mtdoops: Add further error return code checking
  [MTD] [NOR] Test devtype, not definition in flash_probe(), drivers/mtd/devices/lart.c
  ...

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (112 commits)
  ACPI: fix build warning
  Revert "cpuidle: build fix for non-x86"
  ACPI: update intrd DSDT override console messages
  ACPI: update DSDT override documentation
  ACPI: Add "acpi_no_initrd_override" kernel parameter
  ACPI: its a directory not a folder....
  ACPI: misc cleanups
  ACPI: add missing prink prefix strings
  ACPI: cleanup acpi.h
  ACPICA: fix CONFIG_ACPI_DEBUG_FUNC_TRACE build
  ACPI: video: Ignore ACPI video devices that aren't present in hardware
  ACPI: video: reset brightness on resume
  ACPI: video: call ACPI notifier chain for ACPI video notifications
  ACPI: create notifier chain to get hotkey events to graphics driver
  ACPI: video: delete unused display switch on hotkey event code
  ACPI: video: create "brightness_switch_enabled" modparam
  cpuidle: Add a poll_idle method
  ACPI: cpuidle: Support C1 idle time accounting
  ACPI: enable MWAIT for C1 idle
  ACPI: idle: Fix acpi_safe_halt usages and interrupt enabling/disabling
  ...

* 'for-linus' of git://git.o-hand.com/linux-rpurdie-backlight:
  backlight: Add OMAP1 PWL backlight driver
  backlight: Avoid unecessary driver callbacks

* 'for-linus' of git://git.o-hand.com/linux-rpurdie-leds:
  leds: Add HP Jornada 6xx driver
  leds: Remove the now uneeded ixp4xx driver
  leds: Add power LED to the wrap driver
  leds: Fix led-gpio active_low default brightness
  leds: hw acceleration for Clevo mail LED driver
  leds: Add support for hardware accelerated LED flashing
  leds: Standardise LED naming scheme
  leds: Add clevo notebook LED driver

The recently introduced page walker (walk_page_range()) calls pgd_offset with a
const struct mm_struct pointer, causing the following compile warning on m68k:

    mm/pagewalk.c:111: warning: passing argument 1 of 'pgd_offset' discards qualifiers from pointer target type

Make the `mm' parameter of the inline function pgd_offset() const to shut it
up.
Signed-off-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

* 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (22 commits)
  drm: add initial r500 drm support
  radeon: setup the ring buffer fetcher to be less agressive.
  drm: fixup some of the ioctl function exit paths
  drm: the drm really should call pci_set_master..
  i915: Add chipset id for Intel Integrated Graphics Device
  drm: cleanup DRM_DEBUG() parameters
  drm/i915: add support for E7221 chipset
  drm: don't cast a pointer to pointer of list_head
  mga_dma: return 'err' not just zero from mga_do_cleanup_dma()
  drm: add _DRM_DRIVER flag, and re-order unload.
  drm: enable udev node creation
  drm: Make DRM_IOCTL_GET_CLIENT return EINVAL when it can't find client #idx.
  drm: move drm_mem_init to proper place in startup sequence
  drm: call driver load function after initialising AGP
  drm: Fix ioc32 compat layer
  drm: fd.o bug #11895: Only add the AGP base to map offset if the caller didn't.
  i915: add suspend/resume support
  drm: update DRM sysfs support
  drm: Initialize the AGP structure's base address at init rather than enable.
  drm: move two function extern into the correct block
  ...

* git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] Add missing printk levels to e_powersaver
  [CPUFREQ] Fix sparse warning in powernow-k8
  [CPUFREQ] Support Model D parts and newer in e_powersaver
  [CPUFREQ] Powernow-k8: Update to support the latest Turion processors
  [CPUFREQ] fix configuration help message
  [CPUFREQ] powernow-k8 print pstate instead of fid/did for family 10h
  [CPUFREQ] Eliminate cpufreq_userspace scaling_setspeed deadlock
  [CPUFREQ] gx-suspmod.c: use boot_cpu_data instead of current_cpu_data
  [CPUFREQ] fix incorrect comment on show_available_freqs() in freq_table.c
  [CPUFREQ] drivers/cpufreq: Add missing "space"
  [CPUFREQ] arch/x86: Add missing "space"
  [CPUFREQ] Remove pointless Kconfig dependancy

* 'for-2.6.25' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc: (69 commits)
  [POWERPC] Add SPE registers to core dumps
  [POWERPC] Use regset code for compat PTRACE_*REGS* calls
  [POWERPC] Use generic compat_sys_ptrace
  [POWERPC] Use generic compat_ptrace_request
  [POWERPC] Use generic ptrace peekdata/pokedata
  [POWERPC] Use regset code for PTRACE_*REGS* requests
  [POWERPC] Switch to generic compat_binfmt_elf code
  [POWERPC] Switch to using user_regset-based core dumps
  [POWERPC] Add user_regset compat support
  [POWERPC] Add user_regset_view definitions
  [POWERPC] Use user_regset accessors for GPRs
  [POWERPC] ptrace accessors for special regs MSR and TRAP
  [POWERPC] Use user_regset accessors for SPE regs
  [POWERPC] Use user_regset accessors for altivec regs
  [POWERPC] Use user_regset accessors for FP regs
  [POWERPC] mpc52xx: fix compile error introduce when rebasing patch
  [POWERPC] 4xx: PCIe indirect DCR spinlock fix.
  [POWERPC] Add missing native dcr dcr_ind_lock spinlock
  [POWERPC] 4xx: Fix offset value on Warp board
  [POWERPC] 4xx: Add 440EPx Sequoia ehci dts entry
  ...

Add new card (0x1393:0x1143) support added in 1.11 original driver, also
allow rate change in set_serial_info ioctl (as per 1.11 too).
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

(Old) mxser is obsoleted by mxser_new and scheduled for removal on Dec 2007.
Remove it by renaming mxser_new to mxser.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- remove dead MOXA_GET_CONF (always returned -ENXIO)
- remove useless MOXA_GET_CUMAJOR (unused)
- use get/put_user instead of copy_from/to_user for simple types
- cleanup TIOCMIWAIT -- return -ERESTARTSYS on signal, move condition into
  separate function
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Initialize temp structure directly with proper values without first zeroing
it and setting later as suggested by Jan.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Jan Engelhardt <jengelh@computergmbh.de>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Reorder fields to save some memory and code on 64bit due to alignment as
suggested by Jan.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Jan Engelhardt <jengelh@computergmbh.de>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Don't test a pointer against 0. Use NULL instead.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Let the special baudrate processing on the tty layer.  Also remove
set/get_special_rate ioctls introduced in commit
f64c84a1, since it is no longer needed.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Reviewed-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAlan Cox <alan@redhat.com>
Cc: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Both of them may be called directly from the code, don't add special code
and variables and schedule a work for them.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- Cy_EVENT_OPEN_WAKEUP is simple wake_up
- Cy_EVENT_HANGUP is wake_up + tty_hangup, which schedules its own work
- Cy_EVENT_WRITE_WAKEUP is tty_wakeup which may be called directly too
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- tty_hangup schedules a bottomhalf itself, tty_wakeup doesn't need it
- call the CD code (part of work handler previously) directly from the code
  (it wakes somebody up or calls tty_hangup at worse)
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- tqueue is used only for tty_wakeup, call it directly from the code
- tqueue_hangup for tty_hangup, it schedules its own work, use it directly
  too
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

tty_hangup schedules a work for hangup itself, no need to do it in the driver.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

There is no need to schedule a bottomhalf for either of them. One is fast
and the another schedules a bottomhalf itself.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Let compiler decide if the rc_init_drivers function will be inlined and
mark it as __init, because it's called only from __init function.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Don't emit warnings on 64 bit platforms from min(). sizeof() on those
is not uint, neither 2 pointers difference, cast it to uint by min_t in
both cases.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

To not pass ulong address as int parameter, switch it to ulong.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Acked-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

- add KERN_ level to each print
- change some levels appropriately
- add \n at the ends where missing
- change two complex printks into dev_info, where the original info is
  printed automatically
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Don't busy wait for whole 1s when registering some rocket modems. Sleep
instead since we are not in atomic.
Signed-off-by: NJiri Slaby <jirislaby@gmail.com>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Needs the following in order to work correctly on my Inspiron E1705:

Add DMI Product name to i8k for Dell MP061 hardware (Inspiron 9400/E1705)
Signed-off-by: NFrank Sorenson <frank@tuxrocks.com>
Cc: Bradley Smith <bradjsmith@btinternet.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Adds i8k driver to the x86_64 Kconfig.
Signed-off-by: NBradley Smith <bradjsmith@btinternet.com>
Cc: Frank Sorenson <frank@tuxrocks.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Adds #if clause and additional inline assembly so that the driver
builds on x86_64 systems.
Signed-off-by: NBradley Smith <bradjsmith@btinternet.com>
Cc: Frank Sorenson <frank@tuxrocks.com>
Cc: Jeff Garzik <jeff@garzik.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use the architecture specific __cmpxchg_u32 for 32 bits cmpxchg)_local. Else,
use the new generic cmpxchg_local (disables interrupt).
Signed-off-by: NMathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Chris Zankel <chris@zankel.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use the new generic cmpxchg_local (disables interrupt). Also use the generic
cmpxchg as fallback if SMP is not set.
Signed-off-by: NMathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Miles Bader <miles.bader@necel.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use cmpxchg_u32 and cmpxchg_u64 for cmpxchg_local and cmpxchg64_local. For other
type sizes, use the new generic cmpxchg_local (disables interrupt).

Change:
Since the header depends on local_irqsave/local_irqrestore, it must be
included after their declaration.

Actually, being below the
#include <linux/irqflags.h> should be enough, and on sparc64 it is
included at the beginning of system.h.

So it makes sense to move it up for sparc64.
Signed-off-by: NMathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Move cmpxchg and add cmpxchg_local to system.h.
Use the new generic cmpxchg_local (disables interrupt).
Signed-off-by: NMathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: William Lee Irwin III <wli@holomorphy.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>