This patch is from Nathan Lynch <nathanl@austin.ibm.com>.

The NUMA properties of all "possible" cpus are not necessarily available at
boot time on ppc64 LPAR.  Only the properties for present cpus are known.

This patch modifies the ppc64 numa code to map a cpu to its node right before
it is brought up -- this means that secondary cpus are now mapped to their
nodes during smp_init().  Cpus are removed from their nodes after they have
gone offline.

Also some minor cleanups:

- Stash the "minimum common depth" in a global at boot time, so we don't
  have to rediscover it every time something changes.

- Remove unnecessary variable from of_get_associativity() which is accessed
  while possibly uninitialized.

- Remove the cpu portion from dump_numa_topology() since it will show only
  the boot cpu now.  We could display this information from smp_cpus_done() if
  necessary.
Signed-off-by: Nathan Lynch <nathanl@austin.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch just removes more of the infrastructure in the PPC64 iSeries
console driver that is no longer needed since we no longer need to do
copies from user mode in the tty drivers.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch by Kaigai Kohei fixes a bug in the SELinux sidtab code, where we
do a spin_unlock_irq() while nested under another irq lock, which enables
interrupts and allows a deadlock to happen:

  sidtab_set() is called between POLICY_WRLOCK and POLICY_WRUNLOCK in
  services.c:1092.  sidtab_set() uses SIDTAB_LOCK()/SIDTAB_UNLOCK(), but
  SIDTAB_UNLOCK() enables any interruptions because it's defined as
  spin_unlock_irq().  If an interruption occurs between SIDTAB_UNLOCK() and
  POLICY_WRUNLOCK, and interruption context try to hold the POLICY_RDLOCK,
  then a deadlock happen in the result.

The solution is to save & restore flags on the inner lock, per the patch
below.
Signed-off-by: James Morris <jmorris@redhat.com>
Signed-off-by: Stephen Smalley <sds@epoch.ncsc.mil>
Signed-off-by: Kaigai Kohei <kaigai@ak.jp.nec.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

This patch fixes and simplifies locking in the SELiunux netif cache.

An old problem (which I forgot about) is fixed where a netif lookup can be
followed by a preemption, causing a race against sel_netif_put().  Kaigai
Kohei discovered a problem where netif lookups were also not protected
against races with sel_netif_flush().

The code has now been reworked to fix these problems, eliminate the
refcounting and remove atomic operations entirely from the read path
(generally making better use of RCU).  The avc entry ref has been removed
as part of this simplification in anticipation of an RCU scalability patch
which removes them in general.
Signed-off-by: James Morris <jmorris@redhat.com>
Signed-off-by: Stephen Smalley <sds@epoch.ncsc.mil>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Provide a function to get the pageframe number of the nth page at
scatterlist.page.  We cannot just index off scatterlist.page because the
physically-contiguous pages may not be contiguous in mem_map[].
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

CRIS does not demand alignment, so PageAnon's PAGE_MAPPING_ANON bit got
mixed up with the low bit of the struct address_space *mapping pointer.

Patch based on that from Mikael Starvik, but moved the alignment to the
declaration of struct address_space itself, and align to sizeof(long) so
it's well-aligned on all architectures.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

High latency observed when deleting a large video file from tmpfs.

shmem_truncate held info->lock (easily dropped) and nested atomic kmaps
(more awkward to handle, since it's structured to keep them) while scanning
the many pages of its swap vector.  Now be prepared to cond_resched every
64 ops (but scan an empty page in one go).

shmem_free_pages to free a linked list of empty swap vector pages at the
end (cond_resched every 64): could still free them one by one in the main
loop, but this foreshadows scalability changes - which will want to use RCU
on them, only freeing the pages after a grace period.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

CommitLimit was a good addition to /proc/meminfo, but we don't usually show
both what's used and what's free: don't waste lines of screenspace, omit
CommitAvail, let the user do the arithmetic as with all the others.  And in
updating that Documentation, removed the long-gone ReverseMaps.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The sixth "data" field of /proc/$pid/statm was sometimes negative: text is
a subset of shared_vm, and so was subtracted twice from total_vm.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: William Irwin <wli@holomorphy.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The third "shared" field of /proc/$pid/statm in 2.4 was a count of pages in
the mm whose page_count is more than 1 (oddly, including pages shared just
with swapcache).  That's too costly to calculate each time, so 2.6 changed
it to the total file-backed extent.  But Andrea knows apps and users
surprised when (rss - shared) goes negative: we need to provide an rss-like
statistic, close to the 2.4 interpretation.

Something that's quick and easy to maintain accurately is mm->anon_rss, the
count of anonymous pages in the mm.  Then shared = rss - anon_rss gives a
pretty good and meaningful approximation to 2.4's intention: wli confirms
that this will be useful to Oracle too.

Where to show it?  I think it's best to treat this as a bugfix and show it
in the third field of /proc/$pid/statm, after resident, as before - there's
no evidence that the total file-backed extent was found useful.

Albert would like other fields to revert to page counts, but that's a lot
harder: if mprotect can change the category of a page, then it can't be
accounted as simply as this.  Only go that route if real need shown.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: William Irwin <wli@holomorphy.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The procfs shared_vm accounting in do_mmap_pgoff didn't balance with munmap
in the case of shared anonymous: because file comes in NULL, whereas
vm_file gets set at the end by shmem_zero_setup.

Update file; and update vm_flags (a driver is likely to add VM_IO or
VM_RESERVED, modifying reserved_vm); and update pgoff (doesn't affect
procfs accounting, but could affect vma_merge - though at present all
drivers which modify vm_pgoff set a VM_SPECIAL which prevents merging). 
And do that __vm_stat_account before advancing to make_pages_present.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: William Irwin <wli@holomorphy.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

The NUMA policy for shared memory or tmpfs page allocation was protected by
a semaphore.  It helps to improve scalability if we change that to a
spinlock (and there's only one place that needs to drop and reacquire). 
Oh, and don't even bother to get the spinlock while the tree is empty.
Acked-by: Andi Kleen <ak@suse.de>
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

During system boot many probes, etc.  will generate upcalls to userspace
via call_usermodehelper().  This has the effect of calling execve() before
the key subsystem has been initialized, and thus Oopsing on a NULL key_jar
during key_alloc().  Move key_init to security_initcall so that it's called
along with other security initialization routines which have similar
requirements.
Signed-off-by: Chris Wright <chrisw@osdl.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since xfrm_user can be built as a module it's best not to panic
when something goes wrong in xfrm_user_init.  Call me a wimp for
not fixing netlink_kernel_init's return value :)
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch checks the return value of netlink_insert() in
netlink_kernel_create().  It could fail if someone loads
the same module twice for instance.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch fixes 3 problems with the nommu.c support code:

1. export mem_map and vmtruncate (to make them the same as MMU version)

2. update the total_vm usage in mmap routines
   Without this the OOM killer has no "badness" points to rate
   processes on...

   This patch was originally submitted by Giovanni Casoli
   <giovanni.casoli@atengineering.it> for a 2.4.26 kernel.
   I applied it to the 2.6.9 code.

3. provide stub for arch_get_unmapped_area
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Add code to support the apply_relocate_add() function, needed to
support all relocation types of the m68k and Coldfire families.
This code is straight out of the m68k support for this.

Patch was originally from Christian Magnusson
<christian.magnusson@runaware.com>
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

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

Clean up the definition of HZ on m68knommu systems.
There is really only a couple of cases to handle, most
boards use the traditional 100Hz still. Only special
case out those that don't. The current code is just
spaghetti.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

Remove uneeded includes. asm/delay.h is not needed here.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove uneeded includes. asm/delay.h is not needed here.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove uneeded includes. asm/delay.h is not needed here.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove unused shglcore.h include from m68knommu/mm/memory.c
That file will be removed in an upcoming patch.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Remove unused shglcore.h include from m68knommu/mm/init.c
That file will be removed in an upcoming patch.
Signed-off-by: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

Kbuild can't fix up "-I $(src)/.." to the right directory
due to the particular magic transformations it does on -I
arguments. Remove the space, use "-I$(src)/.." and it works.

Hopefully Sam will fix this build issue. 

I think somebody should dress up as scripts/Makefile.lib
for Halloween. Now _that_ would be scary.

Noted by Al Viro, superhacker and endlessly patient person.

Sign up today!

You too can become a superhacker by setting up a cross-compile
environment for all architectures and building "allmodconfig"
trees all day long to find places where others have screwed up.

If an arch Makefile overrides CROSS_COMPILE (e.g. parisc, mips, ...)
then the cc-option call in the main Makefile uses the wrong compiler 
to check for options.
Signed-off-by: Randolph Chung <tausq@debian.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

	FB_INTEL selects AGP_INTEL (i.e. intel-agp.c), which is i386-only.
AGP_INTEL itself depends on X86 && !X86_64 and for a good reason - it doesn't
build elsewhere.

	Added dependency to FB_INTEL itself; as it is allmodconfig had been
broken on everything outside i386.
Signed-off-by: Al Viro <viro@parcelfarce.linux.theplanet.co.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

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

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

Host drivers should now use the scatterlist rather than accessing
the request directly.  This marks the request pointer as being
deprecated so that host driver authors get a chance to fix their
drivers up before we remove this member.

This converts the PXA MCI driver to use the supplied scatter list
rather than accessing the struct request directly, thereby making
this host driver independent of the block layer.

We also clean up the driver to use a consistent driver name for
resources.

Convert MMCI driver to use the scatter list provided by the upper
MMC layers.  This removes all knowledge of struct request from the
MMCI driver.

This is intended to remove block layer knowledge from MMC host
drivers.  Host drivers are expected to use the scatter list for
data transfers rather than accessing the struct request.

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