This implements automatic code patching of memory barriers based
on the CPU capabilities. Normally lock ; addl $0,(%esp) barriers
are used, but these are a bit slow on the Pentium 4.

Linus proposed this a few weeks ago after the support for SSE1/SSE2
barriers was introduced. I now got around to implement it.

The main advantage is that it allows distributors to ship less binary
kernels but still get fast kernels. In particular it avoids the
need of a special Pentium 4 kernel.

The patching code is quite generic and could be used to patch
other instructions (like prefetches or specific other critical
instructions) too.
Thanks to Rusty's in kernel loader it also works seamlessly for modules.

The patching is done before other CPUs start to avoid potential
erratas with self modifying code on SMP systems. It makes no
attempt to automatically handle assymetric systems (an secondary
CPU having less capabilities than the boot CPU). In this
case just boot with "noreplacement"

The VMA merging code can merge vmas which have a ->vm_ops->close() handler.
But this means that per-VMA resources in one of the merged VMAs will not be
freed.

Fix that up by just refusing to merge any VMA's which look "complex": they
have special flags set or they have a ->close handler.

into penguin.transmeta.com:/home/penguin/torvalds/repositories/kernel/linux

Here is a trivial fix for task_prio() in the case MAX_RT_PRIO !=
MAX_USER_RT_PRIO.  In this case, all priorities are skewed by
(MAX_RT_PRIO - MAX_USER_RT_PRIO).

The fix is to subtract the full MAX_RT_PRIO value from p->prio, not just
MAX_USER_RT_PRIO.  This makes sense, as the full priority range is
unrelated to the maximum user value.  Only the real maximum RT value
matters.

This has been in Andrew's tree for awhile, with no issue.  Also, Ingo
acked it.

Ok, now that most drivers have been converted to the new
irqreturn_t, we can remove the fascist type-checks and just
use a regular integer type which has a simpler calling
convention.

With suggestions from Chritoph Hellwig

- Remove MODULE_USE_COUNT macros
- Add owner member to struct tty_ldisc
- Init tty_ldisc at compile time
- make some functions static

- Remove MODULE_USE_COUNT macros
- Add owner member
- Add tiocmget/tiocmset tty callbacks

- Remove MODULE_USE_COUNT macros
- Add owner member
- Add tiocmget/tiocmset tty callbacks

- Remove MODULE_USE_COUNT macros
- Add owner member
- Add tiocmget/tiocmset tty callbacks

- New irq handler prototypes.
- Lots of un-enumurated locking fixes/cleanups (thanks in large part to
  spinlock debug compile options in the kernel).
- Other various trivial fixes.

somebody else.

This converts the Alpha architecture to the new IRQ API.  Tested on
Jensen.

into home.transmeta.com:/home/torvalds/v2.5/linux

Update hotplug and pcmcia drivers to the new IRQ API.

Update the input and joystick drivers to the new IRQ API.

Update ISDN for the new IRQ API.

Update the IRDA drivers to the new IRQ API.

spin_lock() does not take a spinlock_t**

parport_serial is calling a couple of 8250 driver functions with no prototypes
in scope.

Turns out it was calling them with too many args, too.

Update 1394 to the new IRQ API

Update video drivers to the new IRQ API.

Almost none of them compile, so it was done on-spec.

Update a couple of media drivers to the new IRQ API.

Some stuff in there doesn't link btw:

drivers/built-in.o: In function `rds_waitread':
drivers/built-in.o(.text+0x6c152): undefined reference to `aci_port'
drivers/built-in.o: In function `rds_rawwrite':
drivers/built-in.o(.text+0x6c1a3): undefined reference to `aci_port'

Fix up a bunch of char drivers for the new IRQ API.

Update the watchdog drivers to the new IRQ API.

Also, give a couple of irq handlers static scope.

Update the ipmi driver to the new IRQ API

Updates all sound drivers to the new IRQ API.

The patch also fixes

- a return-with-lock-held

- a bunch of warnings and 64-bit bugs (CPU flags must be held in an
  unsigned long)

- jiffies must use unsigned longs

- two functions returning uninitialised values

update rtc driver to new IRQ API

update sym2 for new IRQ API

update drivers/block for new IRQ API.

Affected drivers: atp, de600, de620, eepro, eexpress, lp486e, 
3c589_cs, axnet_cs, fmvj18x_cs, nmclan_cs, pcnet_cs, smc91c92_cs,
xirc2ps_cs, sk98lin, xircom_cb, xircom_tulip_cb, airo, arlan,
netwave_cs, orinoco, ray_cs, wavelan, znet

98% contributed by Andrew Morton

Affected drivers: 3c505 (notably), depca, ni5010, ni52, ni65,
dmfe, winbond-840.

3c505 included quite a few s/int timeout/unsigned long timeout/
changes as well, for proper jiffies comparison typing.

98% contributed by Andrew Morton.

Affected drivers: 3c501, 3c507, 3c509, 3c5154, 3c59x, amd8111e,
at1700, e1000, hp100, lance, smc9194, de2104x, de4x5, tulip,
typhoon.

98% contributed by Andrew Morton.

Affected drivers: 8139cp, 8139too, 82596, b44, cs89x0, dl2k, r8169

Vast majority of changes contributed by Andrew Morton.

morq irqreturn_t,
some s/long flags/unsigned long flags/,
more set_bit-on-something-other-than-long cleanups

Affected drivers: epic100, fealnx, natsemi, ns83820, starfire,
sundance, via-rhine, yellowfin.  (and also include/linux/eeprom.h)

Several changes contributed by Andrew Morton.

Mostly updating to new irqreturn_t, but also includes some needed
SET_MODULE_OWNER and set_bit cleanups as well.

Affects: 8390, dgrs, eepro100, epic100, pcnet32, rcpci45,
sis900, tlan

into redhat.com:/garz/repo/net-drivers-2.5