sfrench/cifs-2.6.git
14 years agopage allocator: do not disable interrupts in free_page_mlock()
Mel Gorman [Tue, 16 Jun 2009 22:32:08 +0000 (15:32 -0700)]
page allocator: do not disable interrupts in free_page_mlock()

free_page_mlock() tests and clears PG_mlocked using locked versions of the
bit operations.  If set, it disables interrupts to update counters and
this happens on every page free even though interrupts are disabled very
shortly afterwards a second time.  This is wasteful.

This patch splits what free_page_mlock() does.  The bit check is still
made.  However, the update of counters is delayed until the interrupts are
disabled and the non-lock version for clearing the bit is used.  One
potential weirdness with this split is that the counters do not get
updated if the bad_page() check is triggered but a system showing bad
pages is getting screwed already.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Acked-by: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: do not call get_pageblock_migratetype() more than necessary
Mel Gorman [Tue, 16 Jun 2009 22:32:07 +0000 (15:32 -0700)]
page allocator: do not call get_pageblock_migratetype() more than necessary

get_pageblock_migratetype() is potentially called twice for every page
free.  Once, when being freed to the pcp lists and once when being freed
back to buddy.  When freeing from the pcp lists, it is known what the
pageblock type was at the time of free so use it rather than rechecking.
In low memory situations under memory pressure, this might skew
anti-fragmentation slightly but the interference is minimal and decisions
that are fragmenting memory are being made anyway.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: inline __rmqueue_fallback()
Mel Gorman [Tue, 16 Jun 2009 22:32:06 +0000 (15:32 -0700)]
page allocator: inline __rmqueue_fallback()

__rmqueue_fallback() is in the slow path but has only one call site.
Because there is only one call-site, this function can then be inlined
without causing text bloat.  On an x86-based config, it made no difference
as the savings were padded out by NOP instructions.  Milage varies but
text will either decrease in size or remain static.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: inline buffered_rmqueue()
Mel Gorman [Tue, 16 Jun 2009 22:32:05 +0000 (15:32 -0700)]
page allocator: inline buffered_rmqueue()

buffered_rmqueue() is in the fast path so inline it.  Because it only has
one call site, this function can then be inlined without causing text
bloat.  On an x86-based config, it made no difference as the savings were
padded out by NOP instructions.  Milage varies but text will either
decrease in size or remain static.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: inline __rmqueue_smallest()
Mel Gorman [Tue, 16 Jun 2009 22:32:04 +0000 (15:32 -0700)]
page allocator: inline __rmqueue_smallest()

Inline __rmqueue_smallest by altering flow very slightly so that there is
only one call site.  Because there is only one call-site, this function
can then be inlined without causing text bloat.  On an x86-based config,
this patch reduces text by 16 bytes.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: remove a branch by assuming __GFP_HIGH == ALLOC_HIGH
Mel Gorman [Tue, 16 Jun 2009 22:32:02 +0000 (15:32 -0700)]
page allocator: remove a branch by assuming __GFP_HIGH == ALLOC_HIGH

Allocations that specify __GFP_HIGH get the ALLOC_HIGH flag.  If these
flags are equal to each other, we can eliminate a branch.

[akpm@linux-foundation.org: Suggested the hack]
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: calculate the alloc_flags for allocation only once
Peter Zijlstra [Tue, 16 Jun 2009 22:32:02 +0000 (15:32 -0700)]
page allocator: calculate the alloc_flags for allocation only once

Factor out the mapping between GFP and alloc_flags only once.  Once
factored out, it only needs to be calculated once but some care must be
taken.

[neilb@suse.de says]
As the test:

-       if (((p->flags & PF_MEMALLOC) || unlikely(test_thread_flag(TIF_MEMDIE)))
-                       && !in_interrupt()) {
-               if (!(gfp_mask & __GFP_NOMEMALLOC)) {

has been replaced with a slightly weaker one:

+       if (alloc_flags & ALLOC_NO_WATERMARKS) {

Without care, this would allow recursion into the allocator via direct
reclaim.  This patch ensures we do not recurse when PF_MEMALLOC is set but
TF_MEMDIE callers are now allowed to directly reclaim where they would
have been prevented in the past.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Neil Brown <neilb@suse.de>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: calculate the migratetype for allocation only once
Mel Gorman [Tue, 16 Jun 2009 22:32:00 +0000 (15:32 -0700)]
page allocator: calculate the migratetype for allocation only once

GFP mask is converted into a migratetype when deciding which pagelist to
take a page from.  However, it is happening multiple times per allocation,
at least once per zone traversed.  Calculate it once.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: calculate the preferred zone for allocation only once
Mel Gorman [Tue, 16 Jun 2009 22:31:59 +0000 (15:31 -0700)]
page allocator: calculate the preferred zone for allocation only once

get_page_from_freelist() can be called multiple times for an allocation.
Part of this calculates the preferred_zone which is the first usable zone
in the zonelist but the zone depends on the GFP flags specified at the
beginning of the allocation call.  This patch calculates preferred_zone
once.  It's safe to do this because if preferred_zone is NULL at the start
of the call, no amount of direct reclaim or other actions will change the
fact the allocation will fail.

[akpm@linux-foundation.org: remove (void) casts]
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: move check for disabled anti-fragmentation out of fastpath
Mel Gorman [Tue, 16 Jun 2009 22:31:58 +0000 (15:31 -0700)]
page allocator: move check for disabled anti-fragmentation out of fastpath

On low-memory systems, anti-fragmentation gets disabled as there is
nothing it can do and it would just incur overhead shuffling pages between
lists constantly.  Currently the check is made in the free page fast path
for every page.  This patch moves it to a slow path.  On machines with low
memory, there will be small amount of additional overhead as pages get
shuffled between lists but it should quickly settle.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: break up the allocator entry point into fast and slow paths
Mel Gorman [Tue, 16 Jun 2009 22:31:57 +0000 (15:31 -0700)]
page allocator: break up the allocator entry point into fast and slow paths

The core of the page allocator is one giant function which allocates
memory on the stack and makes calculations that may not be needed for
every allocation.  This patch breaks up the allocator path into fast and
slow paths for clarity.  Note the slow paths are still inlined but the
entry is marked unlikely.  If they were not inlined, it actally increases
text size to generate the as there is only one call site.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: check only once if the zonelist is suitable for the allocation
Mel Gorman [Tue, 16 Jun 2009 22:31:56 +0000 (15:31 -0700)]
page allocator: check only once if the zonelist is suitable for the allocation

It is possible with __GFP_THISNODE that no zones are suitable.  This patch
makes sure the check is only made once.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: do not check NUMA node ID when the caller knows the node is valid
Mel Gorman [Tue, 16 Jun 2009 22:31:54 +0000 (15:31 -0700)]
page allocator: do not check NUMA node ID when the caller knows the node is valid

Callers of alloc_pages_node() can optionally specify -1 as a node to mean
"allocate from the current node".  However, a number of the callers in
fast paths know for a fact their node is valid.  To avoid a comparison and
branch, this patch adds alloc_pages_exact_node() that only checks the nid
with VM_BUG_ON().  Callers that know their node is valid are then
converted.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: Paul Mundt <lethal@linux-sh.org> [for the SLOB NUMA bits]
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: do not sanity check order in the fast path
Mel Gorman [Tue, 16 Jun 2009 22:31:53 +0000 (15:31 -0700)]
page allocator: do not sanity check order in the fast path

No user of the allocator API should be passing in an order >= MAX_ORDER
but we check for it on each and every allocation.  Delete this check and
make it a VM_BUG_ON check further down the call path.

[akpm@linux-foundation.org: s/VM_BUG_ON/WARN_ON_ONCE/]
Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agopage allocator: replace __alloc_pages_internal() with __alloc_pages_nodemask()
Mel Gorman [Tue, 16 Jun 2009 22:31:52 +0000 (15:31 -0700)]
page allocator: replace __alloc_pages_internal() with __alloc_pages_nodemask()

The start of a large patch series to clean up and optimise the page
allocator.

The performance improvements are in a wide range depending on the exact
machine but the results I've seen so fair are approximately;

kernbench: 0 to  0.12% (elapsed time)
0.49% to  3.20% (sys time)
aim9: -4% to 30% (for page_test and brk_test)
tbench: -1% to  4%
hackbench: -2.5% to  3.45% (mostly within the noise though)
netperf-udp -1.34%  to  4.06% (varies between machines a bit)
netperf-tcp -0.44%  to  5.22% (varies between machines a bit)

I haven't sysbench figures at hand, but previously they were within the
-0.5% to 2% range.

On netperf, the client and server were bound to opposite number CPUs to
maximise the problems with cache line bouncing of the struct pages so I
expect different people to report different results for netperf depending
on their exact machine and how they ran the test (different machines, same
cpus client/server, shared cache but two threads client/server, different
socket client/server etc).

I also measured the vmlinux sizes for a single x86-based config with
CONFIG_DEBUG_INFO enabled but not CONFIG_DEBUG_VM.  The core of the
.config is based on the Debian Lenny kernel config so I expect it to be
reasonably typical.

This patch:

__alloc_pages_internal is the core page allocator function but essentially
it is an alias of __alloc_pages_nodemask.  Naming a publicly available and
exported function "internal" is also a big ugly.  This patch renames
__alloc_pages_internal() to __alloc_pages_nodemask() and deletes the old
nodemask function.

Warning - This patch renames an exported symbol.  No kernel driver is
affected by external drivers calling __alloc_pages_internal() should
change the call to __alloc_pages_nodemask() without any alteration of
parameters.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Dave Hansen <dave@linux.vnet.ibm.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomm: alloc_large_system_hash check order
Hugh Dickins [Tue, 16 Jun 2009 22:31:50 +0000 (15:31 -0700)]
mm: alloc_large_system_hash check order

On an x86_64 with 4GB ram, tcp_init()'s call to alloc_large_system_hash(),
to allocate tcp_hashinfo.ehash, is now triggering an mmotm WARN_ON_ONCE on
order >= MAX_ORDER - it's hoping for order 11.  alloc_large_system_hash()
had better make its own check on the order.

Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: David Miller <davem@davemloft.net>
Cc: Mel Gorman <mel@csn.ul.ie>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Cc: Christoph Lameter <cl@linux.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agocpuset,mm: update tasks' mems_allowed in time
Miao Xie [Tue, 16 Jun 2009 22:31:49 +0000 (15:31 -0700)]
cpuset,mm: update tasks' mems_allowed in time

Fix allocating page cache/slab object on the unallowed node when memory
spread is set by updating tasks' mems_allowed after its cpuset's mems is
changed.

In order to update tasks' mems_allowed in time, we must modify the code of
memory policy.  Because the memory policy is applied in the process's
context originally.  After applying this patch, one task directly
manipulates anothers mems_allowed, and we use alloc_lock in the
task_struct to protect mems_allowed and memory policy of the task.

But in the fast path, we didn't use lock to protect them, because adding a
lock may lead to performance regression.  But if we don't add a lock,the
task might see no nodes when changing cpuset's mems_allowed to some
non-overlapping set.  In order to avoid it, we set all new allowed nodes,
then clear newly disallowed ones.

[lee.schermerhorn@hp.com:
  The rework of mpol_new() to extract the adjusting of the node mask to
  apply cpuset and mpol flags "context" breaks set_mempolicy() and mbind()
  with MPOL_PREFERRED and a NULL nodemask--i.e., explicit local
  allocation.  Fix this by adding the check for MPOL_PREFERRED and empty
  node mask to mpol_new_mpolicy().

  Remove the now unneeded 'nodes = NULL' from mpol_new().

  Note that mpol_new_mempolicy() is always called with a non-NULL
  'nodes' parameter now that it has been removed from mpol_new().
  Therefore, we don't need to test nodes for NULL before testing it for
  'empty'.  However, just to be extra paranoid, add a VM_BUG_ON() to
  verify this assumption.]
[lee.schermerhorn@hp.com:

  I don't think the function name 'mpol_new_mempolicy' is descriptive
  enough to differentiate it from mpol_new().

  This function applies cpuset set context, usually constraining nodes
  to those allowed by the cpuset.  However, when the 'RELATIVE_NODES flag
  is set, it also translates the nodes.  So I settled on
  'mpol_set_nodemask()', because the comment block for mpol_new() mentions
  that we need to call this function to "set nodes".

  Some additional minor line length, whitespace and typo cleanup.]
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Lee Schermerhorn <lee.schermerhorn@hp.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agocpusets: update tasks' page/slab spread flags in time
Miao Xie [Tue, 16 Jun 2009 22:31:47 +0000 (15:31 -0700)]
cpusets: update tasks' page/slab spread flags in time

Fix the bug that the kernel didn't spread page cache/slab object evenly
over all the allowed nodes when spread flags were set by updating tasks'
page/slab spread flags in time.

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agocpusets: restructure the function cpuset_update_task_memory_state()
Miao Xie [Tue, 16 Jun 2009 22:31:46 +0000 (15:31 -0700)]
cpusets: restructure the function cpuset_update_task_memory_state()

The kernel still allocates the page caches on old node after modifying its
cpuset's mems when 'memory_spread_page' was set, or it didn't spread the
page cache evenly over all the nodes that faulting task is allowed to usr
after memory_spread_page was set.  it is caused by the old mem_allowed and
flags of the task, the current kernel doesn't updates them unless some
function invokes cpuset_update_task_memory_state(), it is too late
sometimes.We must update the mem_allowed and the flags of the tasks in
time.

Slab has the same problem.

The following patches fix this bug by updating tasks' mem_allowed and
spread flag after its cpuset's mems or spread flag is changed.

This patch:

Extract a function from cpuset_update_task_memory_state().  It will be
used later for update tasks' page/slab spread flags after its cpuset's
flag is set

Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <cl@linux-foundation.org>
Cc: Paul Menage <menage@google.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Yasunori Goto <y-goto@jp.fujitsu.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomm/page-writeback.c: dirty limit type should be unsigned long
H Hartley Sweeten [Tue, 16 Jun 2009 22:31:44 +0000 (15:31 -0700)]
mm/page-writeback.c: dirty limit type should be unsigned long

get_dirty_limits() calls clip_bdi_dirty_limit() and task_dirty_limit()
with variable pbdi_dirty as one of the arguments.  This variable is an
unsigned long * but both functions expect it to be a long *.  This causes
the following sparse warnings:

  warning: incorrect type in argument 3 (different signedness)
     expected long *pbdi_dirty
     got unsigned long *pbdi_dirty
  warning: incorrect type in argument 2 (different signedness)
     expected long *pdirty
     got unsigned long *pbdi_dirty

Fix the warnings by changing the long * to unsigned long * in both
functions.

Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agovmscan: low order lumpy reclaim also should use PAGEOUT_IO_SYNC
KOSAKI Motohiro [Tue, 16 Jun 2009 22:31:40 +0000 (15:31 -0700)]
vmscan: low order lumpy reclaim also should use PAGEOUT_IO_SYNC

Commit 33c120ed2843090e2bd316de1588b8bf8b96cbde ("more aggressively use
lumpy reclaim") increased how aggressive lumpy reclaim was by isolating
both active and inactive pages for asynchronous lumpy reclaim on
costly-high-order pages and for cheap-high-order when memory pressure is
high.  However, if the system is under heavy pressure and there are dirty
pages, asynchronous IO may not be sufficient to reclaim a suitable page in
time.

This patch causes the caller to enter synchronous lumpy reclaim for
costly-high-order pages and for cheap-high-order pages when under memory
pressure.

Minchan.kim@gmail.com said:

Andy added synchronous lumpy reclaim with
c661b078fd62abe06fd11fab4ac5e4eeafe26b6d.  At that time, lumpy reclaim is
not agressive.  His intension is just for high-order users.(above
PAGE_ALLOC_COSTLY_ORDER).

After some time, Rik added aggressive lumpy reclaim with
33c120ed2843090e2bd316de1588b8bf8b96cbde.  His intention was to do lumpy
reclaim when high-order users and trouble getting a small set of
contiguous pages.

So we also have to add synchronous pageout for small set of contiguous
pages.

Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
Cc: Andy Whitcroft <apw@shadowen.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Minchan Kim <Minchan.kim@gmail.com>
Reviewed-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomm: clean up get_user_pages_fast() documentation
Nick Piggin [Tue, 16 Jun 2009 22:31:39 +0000 (15:31 -0700)]
mm: clean up get_user_pages_fast() documentation

Move more documentation for get_user_pages_fast into the new kerneldoc comment.
Add some comments for get_user_pages as well.

Also, move get_user_pages_fast declaration up to get_user_pages. It wasn't
there initially because it was once a static inline function.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: Nick Piggin <npiggin@suse.de>
Cc: Andy Grover <andy.grover@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: enforce full sync mmap readahead size
Wu Fengguang [Tue, 16 Jun 2009 22:31:38 +0000 (15:31 -0700)]
readahead: enforce full sync mmap readahead size

Now that we do readahead for sequential mmap reads, here is a simple
evaluation of the impacts, and one further optimization.

It's an NFS-root debian desktop system, readahead size = 60 pages.
The numbers are grabbed after a fresh boot into console.

approach        pgmajfault      RA miss ratio   mmap IO count   avg IO size(pages)
   A            383             31.6%           383             11
   B            225             32.4%           390             11
   C            224             32.6%           307             13

case A: mmap sync/async readahead disabled
case B: mmap sync/async readahead enabled, with enforced full async readahead size
case C: mmap sync/async readahead enabled, with enforced full sync/async readahead size
or:
A = vanilla 2.6.30-rc1
B = A plus mmap readahead
C = B plus this patch

The numbers show that
- there are good possibilities for random mmap reads to trigger readahead
- 'pgmajfault' is reduced by 1/3, due to the _async_ nature of readahead
- case C can further reduce IO count by 1/4
- readahead miss ratios are not quite affected

The theory is
- readahead is _good_ for clustered random reads, and can perform
  _better_ than readaround because they could be _async_.
- async readahead size is guaranteed to be larger than readaround
  size, and they are _async_, hence will mostly behave better
However for B
- sync readahead size could be smaller than readaround size, hence may
  make things worse by produce more smaller IOs
which will be fixed by this patch.

Final conclusion:
- mmap readahead reduced major faults by 1/3 and no obvious overheads;
- mmap io can be further reduced by 1/4 with this patch.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: remove redundant test in shrink_readahead_size_eio()
Wu Fengguang [Tue, 16 Jun 2009 22:31:36 +0000 (15:31 -0700)]
readahead: remove redundant test in shrink_readahead_size_eio()

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: introduce context readahead algorithm
Wu Fengguang [Tue, 16 Jun 2009 22:31:36 +0000 (15:31 -0700)]
readahead: introduce context readahead algorithm

Introduce page cache context based readahead algorithm.
This is to better support concurrent read streams in general.

RATIONALE
---------
The current readahead algorithm detects interleaved reads in a _passive_ way.
Given a sequence of interleaved streams 1,1001,2,1002,3,4,1003,5,1004,1005,6,...
By checking for (offset == prev_offset + 1), it will discover the sequentialness
between 3,4 and between 1004,1005, and start doing sequential readahead for the
individual streams since page 4 and page 1005.

The context readahead algorithm guarantees to discover the sequentialness no
matter how the streams are interleaved. For the above example, it will start
sequential readahead since page 2 and 1002.

The trick is to poke for page @offset-1 in the page cache when it has no other
clues on the sequentialness of request @offset: if the current requenst belongs
to a sequential stream, that stream must have accessed page @offset-1 recently,
and the page will still be cached now. So if page @offset-1 is there, we can
take request @offset as a sequential access.

BENEFICIARIES
-------------
- strictly interleaved reads  i.e. 1,1001,2,1002,3,1003,...
  the current readahead will take them as silly random reads;
  the context readahead will take them as two sequential streams.

- cooperative IO processes   i.e. NFS and SCST
  They create a thread pool, farming off (sequential) IO requests to different
  threads which will be performing interleaved IO.

  It was not easy(or possible) to reliably tell from file->f_ra all those
  cooperative processes working on the same sequential stream, since they will
  have different file->f_ra instances. And NFSD's file->f_ra is particularly
  unusable, since their file objects are dynamically created for each request.
  The nfsd does have code trying to restore the f_ra bits, but not satisfactory.

  The new scheme is to detect the sequential pattern via looking up the page
  cache, which provides one single and consistent view of the pages recently
  accessed. That makes sequential detection for cooperative processes possible.

USER REPORT
-----------
Vladislav recommends the addition of context readahead as a result of his SCST
benchmarks. It leads to 6%~40% performance gains in various cases and achieves
equal performance in others.                http://lkml.org/lkml/2009/3/19/239

OVERHEADS
---------
In theory, it introduces one extra page cache lookup per random read.  However
the below benchmark shows context readahead to be slightly faster, wondering..

Randomly reading 200MB amount of data on a sparse file, repeat 20 times for
each block size. The average throughputs are:

                        original ra context ra gain
 4K random reads:  65.561MB/s  65.648MB/s +0.1%
16K random reads: 124.767MB/s 124.951MB/s +0.1%
64K random reads:  162.123MB/s 162.278MB/s +0.1%

Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Tested-by: Vladislav Bolkhovitin <vst@vlnb.net>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: move the random read case to bottom
Wu Fengguang [Tue, 16 Jun 2009 22:31:33 +0000 (15:31 -0700)]
readahead: move the random read case to bottom

Split all readahead cases, and move the random one to bottom.

No behavior changes.

This is to prepare for the introduction of context readahead, and make it
easy for inserting accounting/tracing points for each case.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Vladislav Bolkhovitin <vst@vlnb.net>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoradix-tree: add radix_tree_prev_hole()
Wu Fengguang [Tue, 16 Jun 2009 22:31:32 +0000 (15:31 -0700)]
radix-tree: add radix_tree_prev_hole()

The counterpart of radix_tree_next_hole(). To be used by context readahead.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Vladislav Bolkhovitin <vst@vlnb.net>
Cc: Jens Axboe <jens.axboe@oracle.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: record mmap read-around states in file_ra_state
Wu Fengguang [Tue, 16 Jun 2009 22:31:30 +0000 (15:31 -0700)]
readahead: record mmap read-around states in file_ra_state

Mmap read-around now shares the same code style and data structure with
readahead code.

This also removes do_page_cache_readahead().  Its last user, mmap
read-around, has been changed to call ra_submit().

The no-readahead-if-congested logic is dumped by the way.  Users will be
pretty sensitive about the slow loading of executables.  So it's
unfavorable to disabled mmap read-around on a congested queue.

[akpm@linux-foundation.org: coding-style fixes]
Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Fengguang Wu <wfg@mail.ustc.edu.cn>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: enforce full readahead size on async mmap readahead
Wu Fengguang [Tue, 16 Jun 2009 22:31:29 +0000 (15:31 -0700)]
readahead: enforce full readahead size on async mmap readahead

We need this in one particular case and two more general ones.

Now we do async readahead for sequential mmap reads, and do it with the
help of PG_readahead.  For normal reads, PG_readahead is the sufficient
condition to do a sequential readahead.  But unfortunately, for mmap
reads, there is a tiny nuisance:

[11736.998347] readahead-init0(process: sh/23926, file: sda1/w3m, offset=0:4503599627370495, ra=0+4-3) = 4
[11737.014985] readahead-around(process: w3m/23926, file: sda1/w3m, offset=0:0, ra=290+32-0) = 17
[11737.019488] readahead-around(process: w3m/23926, file: sda1/w3m, offset=0:0, ra=118+32-0) = 32
[11737.024921] readahead-interleaved(process: w3m/23926, file: sda1/w3m, offset=0:2, ra=4+6-6) = 6
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                                                 ~~~~~~~~~~~~~

An unfavorably small readahead.  The original dumb read-around size could
be more efficient.

That happened because ld-linux.so does a read(832) in L1 before mmap(),
which triggers a 4-page readahead, with the second page tagged
PG_readahead.

L0: open("/lib/libc.so.6", O_RDONLY)        = 3
L1: read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\340\342"..., 832) = 832
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
L2: fstat(3, {st_mode=S_IFREG|0755, st_size=1420624, ...}) = 0
L3: mmap(NULL, 3527256, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0x7fac6e51d000
L4: mprotect(0x7fac6e671000, 2097152, PROT_NONE) = 0
L5: mmap(0x7fac6e871000, 20480, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x154000) = 0x7fac6e871000
L6: mmap(0x7fac6e876000, 16984, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x7fac6e876000
L7: close(3)                                = 0

In general, the PG_readahead flag will also be hit in cases

- sequential reads

- clustered random reads

A full readahead size is desirable in both cases.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: sequential mmap readahead
Wu Fengguang [Tue, 16 Jun 2009 22:31:28 +0000 (15:31 -0700)]
readahead: sequential mmap readahead

Auto-detect sequential mmap reads and do readahead for them.

The sequential mmap readahead will be triggered when
- sync readahead: it's a major fault and (prev_offset == offset-1);
- async readahead: minor fault on PG_readahead page with valid readahead state.

The benefits of doing readahead instead of read-around:
- less I/O wait thanks to async readahead
- double real I/O size and no more cache hits

The single stream case is improved a little.
For 100,000 sequential mmap reads:

                                    user       system    cpu        total
(1-1)  plain -mm, 128KB readaround: 3.224      2.554     48.40%     11.838
(1-2)  plain -mm, 256KB readaround: 3.170      2.392     46.20%     11.976
(2)  patched -mm, 128KB readahead:  3.117      2.448     47.33%     11.607

The patched (2) has smallest total time, since it has no cache hit overheads
and less I/O block time(thanks to async readahead). Here the I/O size
makes no much difference, since there's only one single stream.

Note that (1-1)'s real I/O size is 64KB and (1-2)'s real I/O size is 128KB,
since the half of the read-around pages will be readahead cache hits.

This is going to make _real_ differences for _concurrent_ IO streams.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: clean up and simplify the code for filemap page fault readahead
Linus Torvalds [Tue, 16 Jun 2009 22:31:25 +0000 (15:31 -0700)]
readahead: clean up and simplify the code for filemap page fault readahead

This shouldn't really change behavior all that much, but the single rather
complex function with read-ahead inside a loop etc is broken up into more
manageable pieces.

The behaviour is also less subtle, with the read-ahead being done up-front
rather than inside some subtle loop and thus avoiding the now unnecessary
extra state variables (ie "did_readaround" is gone).

Fengguang: the code split in fact fixed a bug reported by Pavel Levshin:
the PGMAJFAULT accounting used to be bypassed when MADV_RANDOM is set, in
which case the original code will directly jump to no_cached_page reading.

Cc: Pavel Levshin <lpk@581.spb.su>
Cc: <wli@movementarian.org>
Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: remove sync/async readahead call dependency
Wu Fengguang [Tue, 16 Jun 2009 22:31:24 +0000 (15:31 -0700)]
readahead: remove sync/async readahead call dependency

The readahead call scheme is error-prone in that it expects the call sites
to check for async readahead after doing a sync one.  I.e.

if (!page)
page_cache_sync_readahead();
page = find_get_page();
if (page && PageReadahead(page))
page_cache_async_readahead();

This is because PG_readahead could be set by a sync readahead for the
_current_ newly faulted in page, and the readahead code simply expects one
more callback on the same page to start the async readahead.  If the
caller fails to do so, it will miss the PG_readahead bits and never able
to start an async readahead.

Eliminate this insane constraint by piggy-backing the async part into the
current readahead window.

Now if an async readahead should be started immediately after a sync one,
the readahead logic itself will do it.  So the following code becomes
valid: (the 'else' in particular)

if (!page)
page_cache_sync_readahead();
else if (PageReadahead(page))
page_cache_async_readahead();

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: increase interleaved readahead size
Wu Fengguang [Tue, 16 Jun 2009 22:31:23 +0000 (15:31 -0700)]
readahead: increase interleaved readahead size

Make sure interleaved readahead size is larger than request size.  This
also makes the readahead window grow up more quickly.

Reported-by: Xu Chenfeng <xcf@ustc.edu.cn>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: remove one unnecessary radix tree lookup
Wu Fengguang [Tue, 16 Jun 2009 22:31:21 +0000 (15:31 -0700)]
readahead: remove one unnecessary radix tree lookup

(hit_readahead_marker != 0) means the page at @offset is present, so we
can search for non-present page starting from @offset+1.

Reported-by: Xu Chenfeng <xcf@ustc.edu.cn>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: apply max_sane_readahead() limit in ondemand_readahead()
Wu Fengguang [Tue, 16 Jun 2009 22:31:21 +0000 (15:31 -0700)]
readahead: apply max_sane_readahead() limit in ondemand_readahead()

Just in case someone aggressively sets a huge readahead size.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: move max_sane_readahead() calls into force_page_cache_readahead()
Wu Fengguang [Tue, 16 Jun 2009 22:31:20 +0000 (15:31 -0700)]
readahead: move max_sane_readahead() calls into force_page_cache_readahead()

Impact: code simplification.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoreadahead: make mmap_miss an unsigned int
Wu Fengguang [Tue, 16 Jun 2009 22:31:19 +0000 (15:31 -0700)]
readahead: make mmap_miss an unsigned int

This makes the performance impact of possible mmap_miss wrap around to be
temporary and tolerable: i.e.  MMAP_LOTSAMISS=100 extra readarounds.

Otherwise if ever mmap_miss wraps around to negative, it takes INT_MAX
cache misses to bring it back to normal state.  During the time mmap
readaround will be _enabled_ for whatever wild random workload.  That's
almost permanent performance impact.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: Ying Han <yinghan@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agomm: consolidate init_mm definition
Alexey Dobriyan [Tue, 16 Jun 2009 22:31:18 +0000 (15:31 -0700)]
mm: consolidate init_mm definition

* create mm/init-mm.c, move init_mm there
* remove INIT_MM, initialize init_mm with C99 initializer
* unexport init_mm on all arches:

  init_mm is already unexported on x86.

  One strange place is some OMAP driver (drivers/video/omap/) which
  won't build modular, but it's already wants get_vm_area() export.
  Somebody should look there.

[akpm@linux-foundation.org: add missing #includes]
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Mike Frysinger <vapier.adi@gmail.com>
Cc: Americo Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agofirmware_map: fix hang with x86/32bit
Yinghai Lu [Tue, 16 Jun 2009 22:31:16 +0000 (15:31 -0700)]
firmware_map: fix hang with x86/32bit

Addresses http://bugzilla.kernel.org/show_bug.cgi?id=13484

Peer reported:
| The bug is introduced from kernel 2.6.27, if E820 table reserve the memory
| above 4G in 32bit OS(BIOS-e820: 00000000fff80000 - 0000000120000000
| (reserved)), system will report Int 6 error and hang up. The bug is caused by
| the following code in drivers/firmware/memmap.c, the resource_size_t is 32bit
| variable in 32bit OS, the BUG_ON() will be invoked to result in the Int 6
| error. I try the latest 32bit Ubuntu and Fedora distributions, all hit this
| bug.
|======
|static int firmware_map_add_entry(resource_size_t start, resource_size_t end,
|                  const char *type,
|                  struct firmware_map_entry *entry)

and it only happen with CONFIG_PHYS_ADDR_T_64BIT is not set.

it turns out we need to pass u64 instead of resource_size_t for that.

[akpm@linux-foundation.org: add comment]
Reported-and-tested-by: Peer Chen <pchen@nvidia.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agospi: takes size of a pointer to determine the size of the pointed-to type
Roel Kluin [Tue, 16 Jun 2009 22:31:15 +0000 (15:31 -0700)]
spi: takes size of a pointer to determine the size of the pointed-to type

Do not take the size of a pointer to determine the size of the pointed-to
type.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Acked-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Cc: David Brownell <david-b@pacbell.net>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Kumar Gala <galak@gate.crashing.org>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agotime: move PIT_TICK_RATE to linux/timex.h
Arnd Bergmann [Tue, 16 Jun 2009 22:31:12 +0000 (15:31 -0700)]
time: move PIT_TICK_RATE to linux/timex.h

PIT_TICK_RATE is currently defined in four architectures, but in three
different places.  While linux/timex.h is not the perfect place for it, it
is still a reasonable replacement for those drivers that traditionally use
asm/timex.h to get CLOCK_TICK_RATE and expect it to be the PIT frequency.

Note that for Alpha, the actual value changed from 1193182UL to 1193180UL.
 This is unlikely to make a difference, and probably can only improve
accuracy.  There was a discussion on the correct value of CLOCK_TICK_RATE
a few years ago, after which every existing instance was getting changed
to 1193182.  According to the specification, it should be
1193181.818181...

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Len Brown <lenb@kernel.org>
Cc: john stultz <johnstul@us.ibm.com>
Cc: Dmitry Torokhov <dtor@mail.ru>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Mon, 15 Jun 2009 17:38:06 +0000 (10:38 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] fix compile error in arch/ia64/mm/extable.c

14 years agoMerge branch 'timers-for-linus-migration' of git://git.kernel.org/pub/scm/linux/kerne...
Linus Torvalds [Mon, 15 Jun 2009 17:06:19 +0000 (10:06 -0700)]
Merge branch 'timers-for-linus-migration' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-for-linus-migration' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  timers: Logic to move non pinned timers
  timers: /proc/sys sysctl hook to enable timer migration
  timers: Identifying the existing pinned timers
  timers: Framework for identifying pinned timers
  timers: allow deferrable timers for intervals tv2-tv5 to be deferred

Fix up conflicts in kernel/sched.c and kernel/timer.c manually

14 years agoMerge branch 'timers-for-linus-clockevents' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Mon, 15 Jun 2009 16:58:50 +0000 (09:58 -0700)]
Merge branch 'timers-for-linus-clockevents' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-for-linus-clockevents' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  clockevent: export register_device and delta2ns
  clockevents: tick_broadcast_device can become static

14 years agoMerge branch 'timers-for-linus-clocksource' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Mon, 15 Jun 2009 16:58:33 +0000 (09:58 -0700)]
Merge branch 'timers-for-linus-clocksource' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-for-linus-clocksource' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  clocksource: prevent selection of low resolution clocksourse also for nohz=on
  clocksource: sanity check sysfs clocksource changes

14 years agoMerge branch 'timers-for-linus-ntp' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 15 Jun 2009 16:43:24 +0000 (09:43 -0700)]
Merge branch 'timers-for-linus-ntp' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-for-linus-ntp' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  ntp: fix comment typos
  ntp: adjust SHIFT_PLL to improve NTP convergence

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next-2.6
Linus Torvalds [Mon, 15 Jun 2009 16:40:05 +0000 (09:40 -0700)]
Merge git://git./linux/kernel/git/davem/net-next-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next-2.6: (1244 commits)
  pkt_sched: Rename PSCHED_US2NS and PSCHED_NS2US
  ipv4: Fix fib_trie rebalancing
  Bluetooth: Fix issue with uninitialized nsh.type in DTL-1 driver
  Bluetooth: Fix Kconfig issue with RFKILL integration
  PIM-SM: namespace changes
  ipv4: update ARPD help text
  net: use a deferred timer in rt_check_expire
  ieee802154: fix kconfig bool/tristate muckup
  bonding: initialization rework
  bonding: use is_zero_ether_addr
  bonding: network device names are case sensative
  bonding: elminate bad refcount code
  bonding: fix style issues
  bonding: fix destructor
  bonding: remove bonding read/write semaphore
  bonding: initialize before registration
  bonding: bond_create always called with default parameters
  x_tables: Convert printk to pr_err
  netfilter: conntrack: optional reliable conntrack event delivery
  list_nulls: add hlist_nulls_add_head and hlist_nulls_del
  ...

14 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc
Linus Torvalds [Mon, 15 Jun 2009 16:32:52 +0000 (09:32 -0700)]
Merge branch 'merge' of git://git./linux/kernel/git/benh/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (103 commits)
  powerpc: Fix bug in move of altivec code to vector.S
  powerpc: Add support for swiotlb on 32-bit
  powerpc/spufs: Remove unused error path
  powerpc: Fix warning when printing a resource_size_t
  powerpc/xmon: Remove unused variable in xmon.c
  powerpc/pseries: Fix warnings when printing resource_size_t
  powerpc: Shield code specific to 64-bit server processors
  powerpc: Separate PACA fields for server CPUs
  powerpc: Split exception handling out of head_64.S
  powerpc: Introduce CONFIG_PPC_BOOK3S
  powerpc: Move VMX and VSX asm code to vector.S
  powerpc: Set init_bootmem_done on NUMA platforms as well
  powerpc/mm: Fix a AB->BA deadlock scenario with nohash MMU context lock
  powerpc/mm: Fix some SMP issues with MMU context handling
  powerpc: Add PTRACE_SINGLEBLOCK support
  fbdev: Add PLB support and cleanup DCR in xilinxfb driver.
  powerpc/virtex: Add ml510 reference design device tree
  powerpc/virtex: Add Xilinx ML510 reference design support
  powerpc/virtex: refactor intc driver and add support for i8259 cascading
  powerpc/virtex: Add support for Xilinx PCI host bridge
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6
Linus Torvalds [Mon, 15 Jun 2009 16:27:37 +0000 (09:27 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/lrg/voltage-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6:
  regulator/max1586: fix V3 gain calculation integer overflow
  regulator/max1586: support increased V3 voltage range
  regulator: lp3971 - fix driver link error when built-in.
  LP3971 PMIC regulator driver (updated and combined version)
  regulator: remove driver_data direct access of struct device
  regulator: Set MODULE_ALIAS for regulator drivers
  regulator: Support list_voltage for fixed voltage regulator
  regulator: Move regulator drivers to subsys_initcall()
  regulator: build fix for powerpc - renamed show_state
  regulator: add userspace-consumer driver
  Maxim 1586 regulator driver

14 years ago[IA64] fix compile error in arch/ia64/mm/extable.c
Rusty Russell [Sat, 13 Jun 2009 05:20:15 +0000 (14:50 +0930)]
[IA64] fix compile error in arch/ia64/mm/extable.c

ad6561dffa17f17bb68d7207d422c26c381c4313 ("module: trim exception table on init
free.") put a bogus trim_init_extable() function into ia64 which didn't compile.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke...
Linus Torvalds [Mon, 15 Jun 2009 16:13:49 +0000 (09:13 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ryusuke/nilfs2

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2: (22 commits)
  nilfs2: support contiguous lookup of blocks
  nilfs2: add sync_page method to page caches of meta data
  nilfs2: use device's backing_dev_info for btree node caches
  nilfs2: return EBUSY against delete request on snapshot
  nilfs2: modify list of unsupported features in caveats
  nilfs2: enable sync_page method
  nilfs2: set bio unplug flag for the last bio in segment
  nilfs2: allow future expansion of metadata read out via get info ioctl
  NILFS2: Pagecache usage optimization on NILFS2
  nilfs2: remove nilfs_btree_operations from btree mapping
  nilfs2: remove nilfs_direct_operations from direct mapping
  nilfs2: remove bmap pointer operations
  nilfs2: remove useless b_low and b_high fields from nilfs_bmap struct
  nilfs2: remove pointless NULL check of bpop_commit_alloc_ptr function
  nilfs2: move get block functions in bmap.c into btree codes
  nilfs2: remove nilfs_bmap_delete_block
  nilfs2: remove nilfs_bmap_put_block
  nilfs2: remove header file for segment list operations
  nilfs2: eliminate removal list of segments
  nilfs2: add sufile function that can modify multiple segment usages
  ...

14 years agoregulator/max1586: fix V3 gain calculation integer overflow
Philipp Zabel [Thu, 28 May 2009 19:00:03 +0000 (21:00 +0200)]
regulator/max1586: fix V3 gain calculation integer overflow

On Thu, May 28, 2009 at 10:59 AM, Mark Brown <broonie@opensource.wolfsonmicro.com> wrote:
> On Thu, May 28, 2009 at 07:15:16AM +0200, Philipp Zabel wrote:
>> The V3 regulator can be configured with an external resistor
>> connected to the feedback pin (R24 in the data sheet) to
>> increase the voltage range.
>>
>> For example, hx4700 has R24 = 3.32 kOhm to achieve a maximum
>> V3 voltage of 1.55 V which is needed for 624 MHz CPU frequency.
>>
>> Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
>
> Looks good.
>
> Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>

Thanks, but it turns out I hit a 32 bit integer overflow in
the gain calculation. I'd like to mend that with the following
patch. Now max_uV could be increased up to 4.294 V, enough to
charge LiPo cells.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator/max1586: support increased V3 voltage range
Philipp Zabel [Thu, 28 May 2009 05:15:16 +0000 (07:15 +0200)]
regulator/max1586: support increased V3 voltage range

The V3 regulator can be configured with an external resistor
connected to the feedback pin (R24 in the data sheet) to
increase the voltage range.

For example, hx4700 has R24 = 3.32 kOhm to achieve a maximum
V3 voltage of 1.55 V which is needed for 624 MHz CPU frequency.

Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: lp3971 - fix driver link error when built-in.
Liam Girdwood [Tue, 19 May 2009 10:44:37 +0000 (11:44 +0100)]
regulator: lp3971 - fix driver link error when built-in.

lp3971_i2c_remove' referenced in section `.data' of drivers/built-in.o:
defined in discarded section `.devexit.text' of drivers/built-in.o

Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoLP3971 PMIC regulator driver (updated and combined version)
Marek Szyprowski [Tue, 19 May 2009 05:33:55 +0000 (07:33 +0200)]
LP3971 PMIC regulator driver (updated and combined version)

This patch adds regulator drivers for National Semiconductors LP3971 PMIC.
This LP3971 PMIC controller has 3 DC/DC voltage converters and 5 low
drop-out (LDO) regulators. LP3971 PMIC controller uses I2C interface.

Reviewed-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: remove driver_data direct access of struct device
Greg Kroah-Hartman [Thu, 30 Apr 2009 22:21:37 +0000 (15:21 -0700)]
regulator: remove driver_data direct access of struct device

In the near future, the driver core is going to not allow direct access
to the driver_data pointer in struct device.  Instead, the functions
dev_get_drvdata() and dev_set_drvdata() should be used.  These functions
have been around since the beginning, so are backwards compatible with
all older kernel versions.

Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: Set MODULE_ALIAS for regulator drivers
Mark Brown [Tue, 28 Apr 2009 10:13:55 +0000 (11:13 +0100)]
regulator: Set MODULE_ALIAS for regulator drivers

Several of the regulator drivers didn't have MODULE_ALIAS so couldn't be
auto loaded. Add the MODULE_ALIAS in case they do get built as modules.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: Support list_voltage for fixed voltage regulator
Mark Brown [Tue, 28 Apr 2009 10:13:54 +0000 (11:13 +0100)]
regulator: Support list_voltage for fixed voltage regulator

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: Move regulator drivers to subsys_initcall()
Mark Brown [Mon, 27 Apr 2009 17:21:18 +0000 (18:21 +0100)]
regulator: Move regulator drivers to subsys_initcall()

Regulators need to be available early in init in order to allow them
to be available for consumers when requested. This is generally done
by registering them at subsys_initcall() time but not all regulator
drivers have done that. Convert these drivers to do so in order to
mimimise future support.

Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Mike Rapoport <mike@compulab.co.il>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: build fix for powerpc - renamed show_state
Liam Girdwood [Tue, 28 Apr 2009 11:01:16 +0000 (12:01 +0100)]
regulator: build fix for powerpc - renamed show_state

This patch fixes the follwing build failure on powerpc:-

> Today's linux-next build (powerpc allyesconfig) failed like this:
>
> drivers/regulator/userspace-consumer.c:43: error: conflicting types
> for 'show_state'
> include/linux/sched.h:273: note: previous definition of 'show_state'
> was here
>
> Caused by commit 5defa2bce704ca4151cfe24e4297aa7797cafd22 ("regulator:
> add userspace-consumer driver") which I have reverted for today.

Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoregulator: add userspace-consumer driver
Mike Rapoport [Sun, 26 Apr 2009 13:49:39 +0000 (16:49 +0300)]
regulator: add userspace-consumer driver

The userspace-consumer driver allows control of voltage and current
regulator state from userspace. This is required for fine-grained
power management of devices that are completely controller by userspace
applications, e.g. a GPS transciever connected to a serial port.

Signed-off-by: Mike Rapoport <mike@compulab.co.il>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoMaxim 1586 regulator driver
Robert Jarzmik [Thu, 23 Apr 2009 18:10:43 +0000 (20:10 +0200)]
Maxim 1586 regulator driver

The Maxim 1586 regulator is a voltage regulator with 2
voltage outputs, specially suitable for Marvell PXA
chips. One output is in the range of required VCC_CORE by
the PXA27x chips, the other in the VCC_USIM required as well
by PXA27x chips.

The chip is controlled through the I2C bus.

Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
14 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/torvalds/linux-2.6
David S. Miller [Mon, 15 Jun 2009 10:02:23 +0000 (03:02 -0700)]
Merge branch 'master' of /linux/kernel/git/torvalds/linux-2.6

Conflicts:
Documentation/feature-removal-schedule.txt
drivers/scsi/fcoe/fcoe.c
net/core/drop_monitor.c
net/core/net-traces.c

14 years agopkt_sched: Rename PSCHED_US2NS and PSCHED_NS2US
Jarek Poplawski [Mon, 15 Jun 2009 09:31:47 +0000 (02:31 -0700)]
pkt_sched: Rename PSCHED_US2NS and PSCHED_NS2US

Let's use TICKS instead of US, so PSCHED_TICKS2NS and PSCHED_NS2TICKS
(like in PSCHED_TICKS_PER_SEC already) to avoid misleading.

Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv4: Fix fib_trie rebalancing
Jarek Poplawski [Mon, 15 Jun 2009 09:31:29 +0000 (02:31 -0700)]
ipv4: Fix fib_trie rebalancing

While doing trie_rebalance(): resize(), inflate(), halve() RCU free
tnodes before updating their parents. It depends on RCU delaying the
real destruction, but if RCU readers start after call_rcu() and before
parent update they could access freed memory.

It is currently prevented with preempt_disable() on the update side,
but it's not safe, except maybe classic RCU, plus it conflicts with
memory allocations with GFP_KERNEL flag used from these functions.

This patch explicitly delays freeing of tnodes by adding them to the
list, which is flushed after the update is finished.

Reported-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Jarek Poplawski <jarkao2@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoramfs: ignore unknown mount options
Mike Frysinger [Sun, 14 Jun 2009 21:56:48 +0000 (22:56 +0100)]
ramfs: ignore unknown mount options

On systems where CONFIG_SHMEM is disabled, mounting tmpfs filesystems can
fail when tmpfs options are used.  This is because tmpfs creates a small
wrapper around ramfs which rejects unknown options, and ramfs itself only
supports a tiny subset of what tmpfs supports.  This makes it pretty hard
to use the same userspace systems across different configuration systems.
As such, ramfs should ignore the tmpfs options when tmpfs is merely a
wrapper around ramfs.

This used to work before commit c3b1b1cbf0 as previously, ramfs would
ignore all options.  But now, we get:
ramfs: bad mount option: size=10M
mount: mounting mdev on /dev failed: Invalid argument

Another option might be to restore the previous behavior, where ramfs
simply ignored all unknown mount options ... which is what Hugh prefers.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Acked-by: Matt Mackall <mpm@selenic.com>
Acked-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild...
Linus Torvalds [Sun, 14 Jun 2009 21:12:18 +0000 (14:12 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/sam/kbuild-next

* 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-next: (53 commits)
  .gitignore: ignore *.lzma files
  kbuild: add generic --set-str option to scripts/config
  kbuild: simplify argument loop in scripts/config
  kbuild: handle non-existing options in scripts/config
  kallsyms: generalize text region handling
  kallsyms: support kernel symbols in Blackfin on-chip memory
  documentation: make version fix
  kbuild: fix a compile warning
  gitignore: Add GNU GLOBAL files to top .gitignore
  kbuild: fix delay in setlocalversion on readonly source
  README: fix misleading pointer to the defconf directory
  vmlinux.lds.h update
  kernel-doc: cleanup perl script
  Improve vmlinux.lds.h support for arch specific linker scripts
  kbuild: fix headers_exports with boolean expression
  kbuild/headers_check: refine extern check
  kbuild: fix "Argument list too long" error for "make headers_check",
  ignore *.patch files
  Remove bashisms from scripts
  menu: fix embedded menu presentation
  ...

14 years ago.gitignore: ignore *.lzma files
Arne Janbu [Wed, 10 Jun 2009 16:25:10 +0000 (18:25 +0200)]
.gitignore: ignore *.lzma files

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Sun, 14 Jun 2009 20:53:22 +0000 (13:53 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  mlx4_core: Don't double-free IRQs when falling back from MSI-X to INTx
  IB/mthca: Don't double-free IRQs when falling back from MSI-X to INTx
  IB/mlx4: Add strong ordering to local inval and fast reg work requests
  IB/ehca: Remove superfluous bitmasks from QP control block
  RDMA/cxgb3: Limit fast register size based on T3 limitations
  RDMA/cxgb3: Report correct port state and MTU
  mlx4_core: Add module parameter for number of MTTs per segment
  IB/mthca: Add module parameter for number of MTTs per segment
  RDMA/nes: Fix off-by-one bugs in reset_adapter_ne020() and init_serdes()
  infiniband: Remove void casts
  IB/ehca: Increment version number
  IB/ehca: Remove unnecessary memory operations for userspace queue pairs
  IB/ehca: Fall back to vmalloc() for big allocations
  IB/ehca: Replace vmalloc() with kmalloc() for queue allocation

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jaswinder/headers-check-2.6
Linus Torvalds [Sun, 14 Jun 2009 20:52:53 +0000 (13:52 -0700)]
Merge git://git./linux/kernel/git/jaswinder/headers-check-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jaswinder/headers-check-2.6:
  headers_check fix: mn10300, setup.h
  headers_check fix: mn10300, ptrace.h

14 years agofusion: fix recent kernel-doc problems
Randy Dunlap [Sun, 14 Jun 2009 02:37:18 +0000 (19:37 -0700)]
fusion: fix recent kernel-doc problems

Fix recent fusion driver kernel-doc fatal error and warnings.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Eric.Moore@lsi.com
Cc: support@lsi.com
Cc: DL-MPTFusionLinux@lsi.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agokeyboard: advertise KT_DEAD2 extended diacriticals
Samuel Thibault [Sat, 13 Jun 2009 12:52:33 +0000 (14:52 +0200)]
keyboard: advertise KT_DEAD2 extended diacriticals

In addition to KT_DEAD which has limited support for diacriticals,
there is KT_DEAD2 that can support 256 criticals, so let's advertise
it in <linux/keyboard.h>.

This lets userland know abut the drivers/char/keyboard.c function
k_dead2, which supports more than the few trivial ones that k_dead
supports.

Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agokbuild: add generic --set-str option to scripts/config
Michal Marek [Mon, 25 May 2009 14:43:27 +0000 (16:43 +0200)]
kbuild: add generic --set-str option to scripts/config

Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agokbuild: simplify argument loop in scripts/config
Michal Marek [Mon, 25 May 2009 14:43:25 +0000 (16:43 +0200)]
kbuild: simplify argument loop in scripts/config

Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agokbuild: handle non-existing options in scripts/config
Michal Marek [Sun, 14 Jun 2009 20:48:07 +0000 (22:48 +0200)]
kbuild: handle non-existing options in scripts/config

If an option does not exist in .config, set it at the end of the file.

Signed-off-by: Michal Marek <mmarek@suse.cz>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc
Linus Torvalds [Sun, 14 Jun 2009 20:46:57 +0000 (13:46 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/drzeus/mmc

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc: (25 commits)
  atmel-mci: add MCI2 register definitions
  atmel-mci: Integrate AT91 specific definition in header file
  tmio_mmc: allow compilation for ASIC3
  mmc_block: do not DMA to stack
  sdhci: Print ADMA status and pointer on debug
  tmio_mmc: fix clock setup
  tmio_mmc: map SD control registers after enabling the MFD cell
  tmio_mmc: correct probe return value for num_resources != 3
  tmio_mmc: don't use set_irq_type
  tmio_mmc: add bus_shift support
  MFD,mmc: tmio_mmc: make HCLK configurable
  mmc_spi: don't use EINVAL for possible transmission errors
  cb710: more cleanup for the DEBUG case.
  sdhci: platform driver for SDHCI
  mxcmmc: remove frequency workaround
  cb710: handle DEBUG define in Makefile
  cb710: add missing parenthesis
  cb710: fix printk format string
  mmc: Driver for CB710/720 memory card reader (MMC part)
  pxamci: add regulator support.
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial
Linus Torvalds [Sun, 14 Jun 2009 20:46:25 +0000 (13:46 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/trivial

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial: (31 commits)
  trivial: remove the trivial patch monkey's name from SubmittingPatches
  trivial: Fix a typo in comment of addrconf_dad_start()
  trivial: usb: fix missing space typo in doc
  trivial: pci hotplug: adding __init/__exit macros to sgi_hotplug
  trivial: Remove the hyphen from git commands
  trivial: fix ETIMEOUT -> ETIMEDOUT typos
  trivial: Kconfig: .ko is normally not included in module names
  trivial: SubmittingPatches: fix typo
  trivial: Documentation/dell_rbu.txt: fix typos
  trivial: Fix Pavel's address in MAINTAINERS
  trivial: ftrace:fix description of trace directory
  trivial: unnecessary (void*) cast removal in sound/oss/msnd.c
  trivial: input/misc: Fix typo in Kconfig
  trivial: fix grammo in bus_for_each_dev() kerneldoc
  trivial: rbtree.txt: fix rb_entry() parameters in sample code
  trivial: spelling fix in ppc code comments
  trivial: fix typo in bio_alloc kernel doc
  trivial: Documentation/rbtree.txt: cleanup kerneldoc of rbtree.txt
  trivial: Miscellaneous documentation typo fixes
  trivial: fix typo milisecond/millisecond for documentation and source comments.
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Sun, 14 Jun 2009 20:45:49 +0000 (13:45 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: fix inverted wheel for bluetooth version of apple mighty mouse
  HID: no more reinitializtion is needed in post_reset
  HID: hidraw -- fix comment about accepted devices
  HID: Multitouch support for the N-Trig touchscreen
  HID: add new multitouch and digitizer contants
  HID: autocentering support for Logitech Force 3D Pro
  HID: fix hid-ff drivers so that devices work even without ff support
  HID: force feedback support for SmartJoy PLUS PS2/USB adapter
  HID: Wacom Graphire Bluetooth driver
  HID: autocentering support for Logitech G25 Racing Wheel

14 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Sun, 14 Jun 2009 20:45:05 +0000 (13:45 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 5545/2: add flush_kernel_dcache_page() for ARM

14 years agokallsyms: generalize text region handling
Mike Frysinger [Mon, 8 Jun 2009 23:12:13 +0000 (19:12 -0400)]
kallsyms: generalize text region handling

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agokallsyms: support kernel symbols in Blackfin on-chip memory
Robin Getz [Mon, 10 Jul 2006 06:25:40 +0000 (06:25 +0000)]
kallsyms: support kernel symbols in Blackfin on-chip memory

The Blackfin arch has a discontiguous .text layout due to having on-chip
instruction memory and no virtual memory support.  As such, we need to
add explicit checks for these additional .text regions.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agoMerge branch 'for-linus' of master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Sun, 14 Jun 2009 20:42:43 +0000 (13:42 -0700)]
Merge branch 'for-linus' of /home/rmk/linux-2.6-arm

* 'for-linus' of master.kernel.org:/home/rmk/linux-2.6-arm: (417 commits)
  MAINTAINERS: EB110ATX is not ebsa110
  MAINTAINERS: update Eric Miao's email address and status
  fb: add support of LCD display controller on pxa168/910 (base layer)
  [ARM] 5552/1: ep93xx get_uart_rate(): use EP93XX_SYSCON_PWRCNT and EP93XX_SYSCON_PWRCN
  [ARM] pxa/sharpsl_pm: zaurus needs generic pxa suspend/resume routines
  [ARM] 5544/1: Trust PrimeCell resource sizes
  [ARM] pxa/sharpsl_pm: cleanup of gpio-related code.
  [ARM] pxa/sharpsl_pm: drop set_irq_type calls
  [ARM] pxa/sharpsl_pm: merge pxa-specific code into generic one
  [ARM] pxa/sharpsl_pm: merge the two sharpsl_pm.c since it's now pxa specific
  [ARM] sa1100: remove unused collie_pm.c
  [ARM] pxa: fix the conflicting non-static declarations of global_gpios[]
  [ARM] 5550/1: Add default configure file for w90p910 platform
  [ARM] 5549/1: Add clock api for w90p910 platform.
  [ARM] 5548/1: Add gpio api for w90p910 platform
  [ARM] 5551/1: Add multi-function pin api for w90p910 platform.
  [ARM] Make ARM_VIC_NR depend on ARM_VIC
  [ARM] 5546/1: ARM PL022 SSP/SPI driver v3
  ARM: OMAP4: SMP: Update defconfig for OMAP4430
  ARM: OMAP4: SMP: Enable SMP support for OMAP4430
  ...

14 years agodocumentation: make version fix
Adam Lackorzynski [Sun, 14 Jun 2009 20:38:59 +0000 (22:38 +0200)]
documentation: make version fix

The Makefiles in the build directories use the internal make variable
MAKEFILE_LIST which is available from make 3.80 only.  (The patch would be
valid back to 2.6.25)

Signed-off-by: Adam Lackorzynski <adam@os.inf.tu-dresden.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agokbuild: fix a compile warning
Amerigo Wang [Wed, 10 Jun 2009 19:48:23 +0000 (12:48 -0700)]
kbuild: fix a compile warning

gcc-4.4.1:

 HOSTCC  scripts/basic/fixdep
scripts/basic/fixdep.c: In function 'traps':
scripts/basic/fixdep.c:377: warning: dereferencing type-punned pointer will break strict-aliasing rules
scripts/basic/fixdep.c:379: warning: dereferencing type-punned pointer will break strict-aliasing rules

(Apparently -fno-strict-aliasing will fix this too)

Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agoMerge branches 'cxgb3', 'ehca', 'misc', 'mlx4', 'mthca' and 'nes' into for-linus
Roland Dreier [Sun, 14 Jun 2009 20:31:19 +0000 (13:31 -0700)]
Merge branches 'cxgb3', 'ehca', 'misc', 'mlx4', 'mthca' and 'nes' into for-linus

14 years agomlx4_core: Don't double-free IRQs when falling back from MSI-X to INTx
Roland Dreier [Sun, 14 Jun 2009 20:30:45 +0000 (13:30 -0700)]
mlx4_core: Don't double-free IRQs when falling back from MSI-X to INTx

When both MSI-X and legacy INTx fail to generate an interrupt, the
driver frees the MSI-X interrupts twice.  Fix this by clearing the
have_irq flag for the MSI-X interrupts when they are freed the first
time.  This is the same bug that was reported in ib_mthca by Yinghai
Lu <yhlu.kernel@gmail.com>.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
14 years agogitignore: Add GNU GLOBAL files to top .gitignore
Jani Nikula [Thu, 11 Jun 2009 09:21:47 +0000 (12:21 +0300)]
gitignore: Add GNU GLOBAL files to top .gitignore

Ignore GPATH, GRTAGS, GSYMS, and GTAGS generated by GNU GLOBAL.

Signed-off-by: Jani Nikula <ext-jani.1.nikula@nokia.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
14 years agokbuild: fix delay in setlocalversion on readonly source
Nico Schottelius [Fri, 12 Jun 2009 07:59:52 +0000 (09:59 +0200)]
kbuild: fix delay in setlocalversion on readonly source

Do not update index on read only media.
Idea published by Christian Kujau <lists@nerdbynature.de>.

Cc: Nico Schottelius <nico@ikn.schottelius.org>
Cc: Christian Kujau <lists@nerdbynature.de>
14 years agoREADME: fix misleading pointer to the defconf directory
Patrick Ringl [Fri, 12 Jun 2009 11:58:36 +0000 (13:58 +0200)]
README: fix misleading pointer to the defconf directory

Signed-off-by: Patrick Ringl <patrick_@freenet.de>
Cc: Arnd Bergmann <arnd@arndb.de>
14 years agovmlinux.lds.h update
Sam Ravnborg [Sun, 14 Jun 2009 20:10:41 +0000 (22:10 +0200)]
vmlinux.lds.h update

Updated after review by Tim Abbott.
- Use HEAD_TEXT_SECTION
- Drop use of section-names.h and delete file
- Introduce EXIT_CALL

Deleting section-names.h required a few simple
updates of init.h

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Tim Abbott <tabbott@ksplice.com>
14 years agoBluetooth: Fix issue with uninitialized nsh.type in DTL-1 driver
Marcel Holtmann [Sun, 14 Jun 2009 13:24:44 +0000 (15:24 +0200)]
Bluetooth: Fix issue with uninitialized nsh.type in DTL-1 driver

The current build shows a warning with the DTL-1 driver:

  CC [M]  drivers/bluetooth/dtl1_cs.o
drivers/bluetooth/dtl1_cs.c: In function ‘dtl1_hci_send_frame’:
drivers/bluetooth/dtl1_cs.c:396: warning: ‘nsh.type’ may be used uninitialized in this function

Fix this by adding a proper error for unknown packet types.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
14 years agoBluetooth: Fix Kconfig issue with RFKILL integration
Marcel Holtmann [Sun, 14 Jun 2009 13:21:25 +0000 (15:21 +0200)]
Bluetooth: Fix Kconfig issue with RFKILL integration

Since the re-write of the RFKILL subsystem it is no longer good to just
select RFKILL, but it is important to add a proper depends on rule.

Based on a report by Alexander Beregalov <a.beregalov@gmail.com>

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
14 years ago[ARM] 5545/2: add flush_kernel_dcache_page() for ARM
Nicolas Pitre [Fri, 12 Jun 2009 02:09:29 +0000 (03:09 +0100)]
[ARM] 5545/2: add flush_kernel_dcache_page() for ARM

Without this, the default implementation is a no op which is completely
wrong with a VIVT cache, and usage of sg_copy_buffer() produces
unpredictable results.

Tested-by: Sebastian Andrzej Siewior <bigeasy@breakpoint.cc>
CC: stable@kernel.org
Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
14 years agoPIM-SM: namespace changes
Tom Goff [Sun, 14 Jun 2009 10:16:13 +0000 (03:16 -0700)]
PIM-SM: namespace changes

IPv4:
  - make PIM register vifs netns local
  - set the netns when a PIM register vif is created
  - make PIM available in all network namespaces (if CONFIG_IP_PIMSM_V2)
    by adding the protocol handler when multicast routing is initialized

IPv6:
  - make PIM register vifs netns local
  - make PIM available in all network namespaces (if CONFIG_IPV6_PIMSM_V2)
    by adding the protocol handler when multicast routing is initialized

Signed-off-by: Tom Goff <thomas.goff@boeing.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMAINTAINERS: EB110ATX is not ebsa110
Russell King [Sun, 14 Jun 2009 10:09:29 +0000 (11:09 +0100)]
MAINTAINERS: EB110ATX is not ebsa110

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
14 years agoMerge branch 'u300' into devel
Russell King [Sun, 14 Jun 2009 10:01:44 +0000 (11:01 +0100)]
Merge branch 'u300' into devel

Conflicts:
arch/arm/Makefile
Updates:
arch/arm/mach-u300/core.c
arch/arm/mach-u300/timer.c

14 years agoMerge branch 'stmp' into devel
Russell King [Sun, 14 Jun 2009 10:01:05 +0000 (11:01 +0100)]
Merge branch 'stmp' into devel

14 years agoMerge branch 'for-rmk' of git://git.kernel.org/pub/scm/linux/kernel/git/ycmiao/pxa...
Russell King [Sun, 14 Jun 2009 10:00:16 +0000 (11:00 +0100)]
Merge branch 'for-rmk' of git://git./linux/kernel/git/ycmiao/pxa-linux-2.6 into devel

14 years agoMerge branch 'copy_user' of git://git.marvell.com/orion into devel
Russell King [Sun, 14 Jun 2009 09:59:32 +0000 (10:59 +0100)]
Merge branch 'copy_user' of git://git.marvell.com/orion into devel

14 years agoipv4: update ARPD help text
Timo Teräs [Thu, 11 Jun 2009 19:38:02 +0000 (19:38 +0000)]
ipv4: update ARPD help text

Removed the statements about ARP cache size as this config option does
not affect it. The cache size is controlled by neigh_table gc thresholds.

Remove also expiremental and obsolete markings as the API originally
intended for arp caching is useful for implementing ARP-like protocols
(e.g. NHRP) in user space and has been there for a long enough time.

Signed-off-by: Timo Teras <timo.teras@iki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>