ppc64 tlb flush rework from Paul Mackerras

Instead of doing a double pass of the pagetables, we batch things
up in the pte flush routines and then shoot the batch down in
flush_tlb_pending.

Our page aging was broken, we never flushed entries out of the ppc64
hashtable. We now flush in ptep_test_and_clear_young.

A number of other things were fixed up in the process:

- change ppc64_tlb_batch to per cpu data
- remove some LPAR debug code
- be more careful with ioremap_mm inits
- clean up arch/ppc64/mm/init.c, create tlb.c

Set the ks bit on userspace segments otherwise the kernel can read/write
into userspace mprotected pages.

This allows us to use more than 2Gb of RAM on a PowerMac G5.  You can
also boot with iommu=off to disable it, but then you get back the 2Gb
RAM limitation.

Make IDE advertise only 128 entries of SG table on archs with an IOMMU.

The current IOMMU implementations of pci_map_sg() may produce segments
that don't match the boundary requirements of IDE, thus causing the
driver to break them up.  The BIO is supposed to account for that,
however, it cannot account for a pci_map_sg producing a segment of the
requested size, but with incorrect alignement, thus we may still try to
break up the list in more entries than is supported by the HW.

A similar fix already went in libata.  The "real" long term fix will be
to move the boundary requirements to struct device so that pci_map_sg()
can respect them when producing the sglist.  In the meantime, this
band-aid works around the problem.

Lots of things renamed, sillicaps killed, stuffs moved around and common
code properly extracted from implementation specific code, new
allocator, etc...  The code is overall a lot simpler, faster, less prone
to fail, and a lot more manageable.  I didn't use "bk mv", there is no
need to keep the old history attached to the new file.

This adds a hack to /dev/mem (along with the other ones already there)
to prevent mapping cacheable of the IO hole.

Without this, XFree blows up on machines with enough memory to go past
the IO hole.  It also tries to prevent memory from beeing mapped
uncached.  Cache paradoxes are evil and can kill the CPU.

The necessary page_is_ram() call was added by the previous patch doing
the proper IO hole accounting.

On machines with an IO hole (like Apple G5 with more than 2Gb of RAM,
but also some pSeries) we failed to properly account for the real
amount of physical RAM and inform the zone allocator of our hole size.

During the process, I included Anton slaughtering of the guard page we
had in the first 256Mb kernel segment, thus allowing this segment to be
mapped with large pages as it should be.

This removes the unused common_regs_m6 and add one more register to be
cleared, according to latest XFree code from ATI.

Fix a stupid warning in PowerMac SMP build on ppc64

Fix the build.

Pass me a brown paper bag.

into ppc970.osdl.org:/home/torvalds/v2.5/linux

into ppc970.osdl.org:/home/torvalds/v2.5/linux

And also make a note in ata_piix.c so this is not forgotten
in the future.

Agreeing with Ben H's recommendation, I reduce the max-hw-segments
number to 128, to match max-phys-segments number.  This shouldn't
affect performance because 128*64K is far above the max transfer size
for most current IDE devices.  Even 128*4K is OK.

From: James Simmons <jsimmons@infradead.org>

This patch creates a separate cursor image drawing region and regular drawing
region.

Avoid cpp clash with the new MODULE_VERSION

From: Rusty Russell <rusty@au1.ibm.com>

The way it works is that the .mod file contains the name of the module (as
before), but succeeding lines are the constituent parts (assumed to be .c
files, which usually works: if they use MODULE_VERSION in a file for which
this isn't true we'll get a warning).

As we postprocess modules, we look in the .modinfo section for a
"version=", which is placed by the MODULE_VERSION() macro.  This will be of
form "version=<macroarg>" "\0" [24 chars] "\0".  The 24 chars are replaced
by the md4 sum of the .c files and any files they #include using '#include
"file"' which are found in the current directory.  Whitespace is collapsed
outside strings, and comments are ignored for purposes of the sum.

The result is a .modinfo entry such as

	version=1.16ac-rustytest B13E9451C4CA3B89577DEFF



At the kernel summit, various people asked for a MODULE_VERSION macro to
store module strings (for later access through sysfs).  A simple md4 is
needed to identify changes in modules which, inevitably, do not update the
version.  It skips whitespace and comments, and includes #includes which
are in the same dir.

The module versions should be set according to this definition, based on
the RPM one, or CVS Revision tags.  Violators will be shot.

 [<epoch>`:']<version>[`-'<extraversion>]
 <epoch>: A (small) unsigned integer which allows you to start versions
          anew. If not mentioned, it's zero.  eg. "2:1.0" is after
     "1:2.0".
 <version>: The <version> may contain only alphanumerics.
 <extraversion>: Like <version>, but inserted for local
          customizations, eg "rh3" or "rusty1".

Comparison of two versions (assuming same epoch):

Split each into all-digit and all-alphabetical parts.  Compare each one one
at a time: digit parts numerically, alphabetical in ASCII order.  So 0.10
comes after 0.9.

From: NeilBrown <neilb@cse.unsw.edu.au>

From: Bruce Allan <bwa@us.ibm.com>

This requires checking if the entry will fit in the current page and, if it
doesn't, encoding it into the next page, and then copying it back to lie
across the boundary between two pages.

From: NeilBrown <neilb@cse.unsw.edu.au>

make sure sunrpc init routines called before gss init routines.

From: NeilBrown <neilb@cse.unsw.edu.au>

Don't rely on the assumption that
    sizeof(stateid_t) == sizeof(stateid_opaque_t) + 4.

From: NeilBrown <neilb@cse.unsw.edu.au>

Fix out-of-spec error return in attribute decoding.

From: NeilBrown <neilb@cse.unsw.edu.au>

Save replays for LOCK operations, make sure errors are replayed correctly.

From: NeilBrown <neilb@cse.unsw.edu.au>

Fix an out-of-spec error in nfsd4_remove.

From: NeilBrown <neilb@cse.unsw.edu.au>

Fix an out-of-spec readlink error return.

From: NeilBrown <neilb@cse.unsw.edu.au>

The spec requires that the access and deny bits requested in open_downgrade
represent the union of the bits for some subset of the OPENS for the given
open_owner and file.  Enforce that requirement.

From: NeilBrown <neilb@cse.unsw.edu.au>

Corrects an error return for OP_OPEN_CONFIRM.

From: NeilBrown <neilb@cse.unsw.edu.au>

owned by the lockowner clientid.

From: NeilBrown <neilb@cse.unsw.edu.au>

fix an error return for OP_CREATE

From: NeilBrown <neilb@cse.unsw.edu.au>

add OP_ILLEGAL, and fix processing of compounds with out of bounds op
numbers.

From: NeilBrown <neilb@cse.unsw.edu.au>

Implement the nfsv4 RELEASE_LOCKOWNER operation.

From: NeilBrown <neilb@cse.unsw.edu.au>

fh_dup2 duplicates an entire svc_fh structure, so in addition to dget()'ing
the dentry, it also needs to bump the reference count on fh_export.

Moved it out of the header file into nfsfh.c for the simple reason that I
couldn't figure out how to reorganize the headers files right so that the
fh_dup2 could use the definition of svc_fh.  Since fh_dup2 is never actually
used outside of nfs4proc.c, it could just as well be moved there....

From: NeilBrown <neilb@cse.unsw.edu.au>

Remove a comment that is no longer accurate

From: NeilBrown <neilb@cse.unsw.edu.au>

add the bookeeping necessary to remove all locks held by an nfsv4 lockowner
upon CLOSE, or upon state expiration.  calls locks_remove_posix().

replace list_del_init() with list_del on nfsv4 state structures that are
being reaped.

From: NeilBrown <neilb@cse.unsw.edu.au>

unlike v2/v3, nfsv4 returns nfserr_exist in some situations where the
underlying filesystem returns nfserr_isdir, nfserr_notdir.

on rename, nfsv4 returns nfserr_notdir instead of nfserr_symlink.

From: NeilBrown <neilb@cse.unsw.edu.au>

check lock length, return appropriate error

From: NeilBrown <neilb@cse.unsw.edu.au>

unlike v2/v3, nfsv4 returns nfserr_inval when attempting to read, write,
commit or test lock a symlink.  nfsv4 does return nfserr_symlink on lookup
and open, so a simple fix in fh_verify() will not work.

From: NeilBrown <neilb@cse.unsw.edu.au>

From: NeilBrown <neilb@cse.unsw.edu.au>

Call decode_fattr on writable attributes to check for xdr errors, incorrect
utf8, etc.

From: NeilBrown <neilb@cse.unsw.edu.au>

tests if current_fh is the pseudo root for the client and returns
nfserr_noent if so.  need to call exp_pseudoroot because different clients
can have