More C99 initializers fixes in ncpfs. By Art Haas <ahaas@neosoft.com>

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

into kroah.com:/home/greg/linux/BK/gregkh-2.5

- I changed the sector_t thing in max_block to use davem's approach.
  I agree with Anton, but making it explicit doesn't hurt.

- Remove a dead comment in copy_strings.

Old stuff:

- Remove the IO error warning in end_buffer_io_sync().  Failed READA
  attempts trigger it.

- Emit a warning when an ext2 is mounting an ext3 filesystem.

  We have had quite a few problem reports related to this, mainly
  arising from initrd problems.  And mount(8) tends to report the
  fstype from /etc/fstab rather than reporting what has really
  happened.

Fixes some bogosity which I added to max_block():

- `size' doesn't need to be sector_t

- `retval' should not be initialised to "~0UL" because that is
  0x00000000ffffffff with 64-bit sector_t.

- Allocate task_structs with GFP_KERNEL, as discussed.

- Convert the EXPORT_SYMBOL for generic_file_direct_IO() to
  EXPORT_SYMBOL_GPL.  That was only exported as a practicality for the
  raw driver.

- Make the loop thread run balance_dirty_pages() after dirtying the
  backing file.  So it will perform writeback of the backing file when
  dirty memory levels are high.  Export balance_dirty_pages to GPL
  modules for this.

  This makes loop work a lot better - I suspect it broke when callers
  of balance_dirty_pages() started writing back only their own queue.

  There are many page allocation failures under heavy loop writeout.
  Coming from blk_queue_bounce()'s allocation from the page_pool
  mempool.  So...

- Disable page allocation warnings around the initial atomic
  allocation attempt in mempool_alloc() - the one where __GFP_WAIT and
  __GFP_IO were turned off.  That one can easily fail.

- Add some commentary in block_write_full_page()

get_page_state() is showing up on profiles on some big machines.  It is
a quite expensive function and it is being called too often.

The patch replaces the hardwired RATELIMIT_PAGES with a calculated
amount based on the amount of memory in the machine and the number of
CPUs.

Replace the buffer lru spinlock protection with local_irq_disable and a
cross-CPU call to invalidate them.

include/linux/prefetch.h does a strange thing: if the arch doesn't have
the prefectch functions, this header defines no-op version of them and
then defines ARCH_HAS_PREFETCH.  So there's no way for mainline code to
know if the architecture *really* has prefetch instructions.

This information loss is unfortunate.  Examples:

	for (i = 0; i < N; i++)
		prefetch(foo[i]);

   Problem is, if `prefetch' is a no-op, the compiler will still
   generate an empty busy-wait loop.  Which it must do.  We need to
   know the truth about ARCH_HAS_PREFETCH to correctly elide that loop.

  - fixed deadlock

This implements CLONE_VM_RELEASE, which lets the child release the 'user
VM' at mm_release() time.

the attached patch implements the per-CPU thread-structure cache to do
detached exit, if the parent does not want to be notified of child exit
via a signal.

This adds two new clone() flags:

    CLONE_SETTLS => if present then the third clone() syscall parameter
                    is the new TLS.

    CLONE_SETTID => if present then the child TID is written to the
                    address specified by the fourth clone() parameter.

the new parameters are handled in a safe way, clone() returns -EFAULT or
-EINVAL if there's some problem with them.

Here is a trivial bug fix for blkdev_reread_part() in 2.5.31.
Without this fix,  "fdisk" hangs with following messages:

	Calling ioctl() to re-read partition table.

	WARNING: Re-reading the partition table failed with error 16:
			Device or resource busy.
	The kernel still uses the old table.
	The new table will be used at the next reboot.

- teaches the shared "hcd" code to set urb->*_dma whenever the device
  driver didn't, by creating singleshot mappings.

-  teaches the hardware-specific code to
   use urb->*_dma instead of creating mappings.
   (And tells ohci-sa1111 to init its buffer pools.)
   EHCI and UHCI also eliminated duplicated state;
   all the HCDs are now a smidgeon smaller.

Sanity checked by enumerating, including through
a hub, and using a USB Ethernet adapter, with each
of the three host controllers.

Worth noting:  this removes pci_dma_sync_single()
calls from UHCI.  On x86 (and some others) that's
a NOP, but for UHCI on other platforms (rare except
maybe on IA64, as I understand) this anticipates
the upcoming patch to remove interrupt automagic.
(I'll likely submit that after a Linus release that
catches up to your USB tree. :)

This patch to 2.5.31 pushes "Networking options" and
"Network device support" together for all architectures
that have them.

They shouldn't be split apart by Telephony, I2O,
Fusion, etc.

This old __per_cpu_data define wasn't enough if an arch wants to
use the gcc __thread prefix (thread local storage), which needs to
go *before* the type in the definition.  So we have to go for a
DECLARE macro, and while we're there, separate DECLARE and DEFINE,
as definitions of per-cpu data cannot live in modules.  This also
means that accidental direct references to per-cpu variables will
be caught at compile time.

This makes introduces get_cpu_var()/put_cpu_var() which gets a
per-cpu variable and disables preemption, and renames the (unsafe
under preemption) "this_cpu()" macro to __get_cpu_var().  It also
deletes the redundant definitions in linux/smp.h.

As per Andrew Morton's request.

This moves the early ioremap call to after cpu_online_map is initialized.
Note everything is wrapped in clustered_apic_mode, so should be safe.

When you reverted the stringify changes I sent last time, you missed removing
a comment, which is now grossly wrong.  This patch finishes the job.

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

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

Most syscall take a const char * argument in fact, also mark them extern.

This is part 3 of 3 of the genrtc patches.  This is my own slight bit
of work, as well as some work by Randolph Chung. This changes
set_rtc_time(struct *rtc_time) to return an int instead of void.
This was done so that the arch-specific code here could do additional
checks on the time and return an error if needed. This then introduces
include/asm-generic/rtc.h, include/asm-i386/rtc.h and
include/asm-alpha/rtc.h.  include/asm-generic/rtc.h contains the
get_rtc_time and set_rtc_time logic that is in drivers/char/rtc.c and
has been tested on SMP i386.  This also modifies include/asm-ppc/rtc.h
to return -ENODEV if no rtc hardware is present.

Additionally, Dave Jones pointed out to me a place where we might not be
safe when jiffies wraps, so this switches that to time_after().

From Randolph Chung, support for a 64bit kernel and a 32bit userland.

This is part 2 of 3 of the genrtc driver work.  This is the PPC portion
of the patch, which creates include/asm-ppc/rtc.h.  This has been in the
PPC bitkeeper tree for over a month now.

This is the current version of the genrtc driver from the m68k
community.  This is slightly different than the version I have sent
previously in that it has been switched to C99-style initializers,
which was done in the current m68k CVS tree by Geert Uytterhoeven, and
the needed changes to select/compile it in general.  I had previously
asked the m68k community if anyone objected to this being submitted by me,
and I got Richard Zidlicky's (who's at the top of the file) approval, as
well as Geert Uytterhoeven's approval.

i2c update:
 o Support for SMBus 2.0 PEC Packet Error Checking
 o New adapter-i2c-frodo for SA 1110 board
 o New adapter-i2c-rpx for embeded MPC8XX
 o Replace depreciated cli()&sti() with spin_{un}lock_irq()
 o Updated documentation

This patch adds sendfile() support to fatfs.

sys_sendfile() call,

    do_sendfile(out_fd, in_fd, &pos, count, MAX_NON_LFS);
        in_file->f_op->sendfile(out_file, in_file, ppos, count);

But,

ssize_t generic_file_sendfile(struct file *in_file, struct file *out_file,
			      loff_t *ppos, size_t count)

fist arg of generic_file_sendfile() should be for output. This patch
fixes this typo.

There are a number of places where generic_file_write could be used if
it didn't take i_sem by itself or where it doesn't need it:

(1)	XFS needs to do some preparation in it's write file operation
	(most notably taking it's own per-inode locks), after taking
	i_sem, but before calling the guts of generic_file_write.
(2)	blockdevices are happy with parralel get_block() operations,
	and there is nothing i_sem could protect.

This patch introduces a generic_file_write_nolock and makes the good old
generic_file_write a wrapper around.  The block device code is switched
over to it.

The old form of designated initializers are obsolete: we need to
replace them with the ISO C forms before 2.6.  Gcc has always supported
both forms anyway.

These are the completely generic bits (linux/init_task.h and linux/wait.h).

From: Art Haas <ahaas@neosoft.com>

  Here's the latest diffs for the files in include/linux.
  Patches are against 2.5.31.

The following patch from Anton Blanchard fixes a problem with the
sym53c8xx_2 driver in the 2.5 tree.  It assumes that the eh error
handlers are not called with the host lock held, which they are.

The code already copes with the possibility of having
read some of the 4byte length but not all, but can never
get into that situation.  We just need to add a 'return'
for the case where read got some data but not enough.
Either 0 or -EAGAIN could be returned, they both have the
same effect.

Now that we actually do shutdown the socket, we should
say so instead of just saying that we should....