GitHub/LineageOS/android_kernel_motorola_exynos9610.git
14 years agoocfs2/dlm: Increase o2dlm lockres hash size
Sunil Mushran [Wed, 5 May 2010 23:25:08 +0000 (16:25 -0700)]
ocfs2/dlm: Increase o2dlm lockres hash size

Lockres hash size of 16KB is far too small for large filesystems (where we
have hundreds of thousands of lock resources stored in the table).
This patch increases it to 128KB.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Make ocfs2_extend_trans() really extend.
Tao Ma [Mon, 26 Apr 2010 06:34:57 +0000 (14:34 +0800)]
ocfs2: Make ocfs2_extend_trans() really extend.

In ocfs2, we use ocfs2_extend_trans() to extend a journal handle's
blocks. But if jbd2_journal_extend() fails, it will only restart
with the the new number of blocks.  This tends to be awkward since
in most cases we want additional reserved blocks. It makes our code
harder to mantain since the caller can't be sure all the original
blocks will not be accessed and dirtied again.  There are 15 callers
of ocfs2_extend_trans() in fs/ocfs2, and 12 of them have to add
h_buffer_credits before they call ocfs2_extend_trans().  This makes
ocfs2_extend_trans() really extend atop the original block count.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2/trivial: Code cleanup for allocation reservation.
Tao Ma [Tue, 6 Apr 2010 08:46:46 +0000 (16:46 +0800)]
ocfs2/trivial: Code cleanup for allocation reservation.

Two tiny cleanup for allocation reservation.
1. Remove some extra codes in ocfs2_local_alloc_find_clear_bits.
2. Remove an unuseful variables in ocfs2_find_resv_lhs.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Acked-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: make ocfs2_adjust_resv_from_alloc simple.
Tao Ma [Thu, 8 Apr 2010 08:33:02 +0000 (16:33 +0800)]
ocfs2: make ocfs2_adjust_resv_from_alloc simple.

When we allocate some bits from the reservation, we always
allocate from the r_start(see ocfs2_resmap_resv_bits).
So there should be no reason to check between r_start
and start. And I don't think we will change this behaviour
later by allocating from some bits after r_start.  Why not make
ocfs2_adjust_resv_from_alloc simple for now?

The only chance we have to adjust the reservation is when we haven't
reached the end. With this patch, the function is more readable.

Note:
btw, this patch also fixes an original bug in the function
which I haven't found before.
if (end < ocfs2_resv_end(resv))
rhs = end - ocfs2_resv_end(resv);
This code is of course buggy. ;)

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Acked-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Make nointr a default mount option
Sunil Mushran [Wed, 14 Apr 2010 01:00:31 +0000 (18:00 -0700)]
ocfs2: Make nointr a default mount option

OCFS2 has never really supported intr. This patch acknowledges this reality
and makes nointr the default mount option. In a later patch, we intend to
support intr.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2/dlm: Make o2dlm domain join/leave messages KERN_NOTICE
Sunil Mushran [Wed, 14 Apr 2010 01:00:30 +0000 (18:00 -0700)]
ocfs2/dlm: Make o2dlm domain join/leave messages KERN_NOTICE

o2dlm join and leave messages are more than informational as they are
required for debugging locking issues. This patch changes them from
KERN_INFO to KERN_NOTICE.

Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoo2net: log socket state changes
Srinivas Eeda [Wed, 31 Mar 2010 21:32:29 +0000 (14:32 -0700)]
o2net: log socket state changes

This patch logs socket state changes that lead to socket shutdown.

Signed-off-by: Srinivas Eeda <srinivas.eeda@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: print node # when tcp fails
Wengang Wang [Tue, 30 Mar 2010 04:09:22 +0000 (12:09 +0800)]
ocfs2: print node # when tcp fails

Print the node number of a peer node if sending it a message failed.

Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Add dir_resv_level mount option
Mark Fasheh [Tue, 6 Apr 2010 01:17:16 +0000 (18:17 -0700)]
ocfs2: Add dir_resv_level mount option

The default behavior for directory reservations stays the same, but we add a
mount option so people can tweak the size of directory reservations
according to their workloads.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: change default reservation window sizes
Mark Fasheh [Tue, 6 Apr 2010 01:17:15 +0000 (18:17 -0700)]
ocfs2: change default reservation window sizes

The default reservation size of 4 (32-bit windows) is a bit too ambitious.
Scale it back to 16 bits (resv_level=2). I have been testing various sizes
on a 4-node cluster which runs a mixed workload that is heavily threaded.
With a 256MB local alloc, I get *roughly* the following levels of average file
fragmentation:

resv_level=0 70%
resv_level=1 21%
resv_level=2 23%
resv_level=3 24%
resv_level=4 60%
resv_level=5 did not test
resv_level=6 60%

resv_level=2 seemed like a good compromise between not letting windows be
too small, but not so big that heavier workloads will immediately suffer
without tuning.

This patch also change the behavior of directory reservations - they now
track file reservations.  The previous compromise of giving directory
windows only 8 bits wound up fragmenting more at some window sizes because
file allocations had smaller unused windows to poach from.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: increase the default size of local alloc windows
Mark Fasheh [Tue, 6 Apr 2010 01:17:14 +0000 (18:17 -0700)]
ocfs2: increase the default size of local alloc windows

I have observed that the current size of 8M gives us pretty poor
fragmentation on multi-threaded workloads which do lots of writes.

Generally, I can increase the size of local alloc windows and observe a
marked decrease in fragmentation, even up and beyond window sizes of 512
megabytes. This makes sense for a couple reasons - larger local alloc means
more room for reservation windows. On multi-node workloads the larger local
alloc helps as well because we don't have to do window slides as often.

Also, I removed the OCFS2_DEFAULT_LOCAL_ALLOC_SIZE constant as it is no
longer used and the comment above it was out of date.

To test fragmentation, I used a workload which launched 4 threads that did
4k writes into a series of about 140 alternating files.

With resv_level=2, and a 4k/4k file system I observed the following average
fragmentation for various localalloc= parameters:

localalloc= avg. fragmentation
8 48
32 16
64 10
120 7

On larger cluster sizes, the difference is more dramatic.

The new default size top out at 256M, which we'll only get for cluster
sizes of 32K and above.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: clean up localalloc mount option size parsing
Mark Fasheh [Tue, 6 Apr 2010 01:17:13 +0000 (18:17 -0700)]
ocfs2: clean up localalloc mount option size parsing

This patch pulls the local alloc sizing code into localalloc.c and provides
a callout to it from ocfs2_fill_super(). Behavior is essentially unchanged
except that I correctly calculate the maximum local alloc size. The old code
in ocfs2_parse_options() calculated the max size as:

ocfs2_local_alloc_size(sb) * 8

which is correct, in bits. Unfortunately though the option passed in is in
megabytes. Ultimately, this bug made no real difference - the shrink code
would catch a too-large size and bring it down to something reasonable.
Still, it's less than efficient as-is.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: remove ocfs2_local_alloc_in_range()
Mark Fasheh [Wed, 17 Mar 2010 04:01:00 +0000 (21:01 -0700)]
ocfs2: remove ocfs2_local_alloc_in_range()

Inodes are always allocated from the global bitmap now so we don't need this
any more. Also, the existing implementation bounces reservations around
needlessly.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
14 years agoocfs2: allocate btree internal block groups from the global bitmap
Mark Fasheh [Wed, 24 Feb 2010 21:34:09 +0000 (13:34 -0800)]
ocfs2: allocate btree internal block groups from the global bitmap

Otherwise, the need for a very large contiguous allocation tends to
wreak havoc on many inode allocation reservations on the local alloc, thus
ruining any chances for contiguousness.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
14 years agoocfs2: use allocation reservations for directory data
Mark Fasheh [Mon, 7 Dec 2009 21:16:07 +0000 (13:16 -0800)]
ocfs2: use allocation reservations for directory data

Use the reservations system for unindexed dir tree allocations. We don't
bother with the indexed tree as reads from it are mostly random anyway.
Directory reservations are marked seperately, to allow the reservations code
a chance to optimize their window sizes. This patch allocates only 8 bits
for directory windows as they generally are not expected to grow as quickly
as file data. Future improvements to dir window sizing can trivially be
made.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
14 years agoocfs2: use allocation reservations during file write
Mark Fasheh [Mon, 7 Dec 2009 21:15:40 +0000 (13:15 -0800)]
ocfs2: use allocation reservations during file write

Add a per-inode reservations structure and pass it through to the
reservations code.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
14 years agoocfs2: allocation reservations
Mark Fasheh [Mon, 7 Dec 2009 21:10:48 +0000 (13:10 -0800)]
ocfs2: allocation reservations

This patch improves Ocfs2 allocation policy by allowing an inode to
reserve a portion of the local alloc bitmap for itself. The reserved
portion (allocation window) is advisory in that other allocation
windows might steal it if the local alloc bitmap becomes
full. Otherwise, the reservations are honored and guaranteed to be
free. When the local alloc window is moved to a different portion of
the bitmap, existing reservations are discarded.

Reservation windows are represented internally by a red-black
tree. Within that tree, each node represents the reservation window of
one inode. An LRU of active reservations is also maintained. When new
data is written, we allocate it from the inodes window. When all bits
in a window are exhausted, we allocate a new one as close to the
previous one as possible. Should we not find free space, an existing
reservation is pulled off the LRU and cannibalized.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
14 years agoocfs2: Make ocfs2_journal_dirty() void.
Joel Becker [Fri, 19 Mar 2010 21:13:52 +0000 (14:13 -0700)]
ocfs2: Make ocfs2_journal_dirty() void.

jbd[2]_journal_dirty_metadata() only returns 0.  It's been returning 0
since before the kernel moved to git.  There is no point in checking
this error.

ocfs2_journal_dirty() has been faithfully returning the status since the
beginning.  All over ocfs2, we have blocks of code checking this can't
fail status.  In the past few years, we've tried to avoid adding these
checks, because they are pointless.  But anyone who looks at our code
assumes they are needed.

Finally, ocfs2_journal_dirty() is made a void function.  All error
checking is removed from other files.  We'll BUG_ON() the status of
jbd2_journal_dirty_metadata() just in case they change it someday.  They
won't.

Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Clear undo bits when local alloc is freed
Mark Fasheh [Fri, 12 Mar 2010 02:31:09 +0000 (18:31 -0800)]
ocfs2: Clear undo bits when local alloc is freed

When the local alloc file changes windows, unused bits are freed back to the
global bitmap. By defnition, those bits can not be in use by any file. Also,
the local alloc will never have been able to allocate those bits if they
were part of a previous truncate. Therefore it makes sense that we should
clear unused local alloc bits in the undo buffer so that they can be used
immediatly.

[ Modified to call it ocfs2_release_clusters() -- Joel ]

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Init meta_ac properly in ocfs2_create_empty_xattr_block.
Tao Ma [Fri, 19 Mar 2010 07:04:24 +0000 (15:04 +0800)]
ocfs2: Init meta_ac properly in ocfs2_create_empty_xattr_block.

You can't store a pointer that you haven't filled in yet and expect it
to work.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Fix the update of name_offset when removing xattrs
Tao Ma [Fri, 19 Mar 2010 07:04:23 +0000 (15:04 +0800)]
ocfs2: Fix the update of name_offset when removing xattrs

When replacing a xattr's value, in some case we wipe its name/value
first and then re-add it. The wipe is done by
ocfs2_xa_block_wipe_namevalue() when the xattr is in the inode or
block. We currently adjust name_offset for all the entries which have
(offset < name_offset). This does not adjust the entrie we're replacing.
Since we are replacing the entry, we don't adjust the total entry count.
When we calculate a new namevalue location, we trust the entries
now-wrong offset in ocfs2_xa_get_free_start().  The solution is to
also adjust the name_offset for the replaced entry, allowing
ocfs2_xa_get_free_start() to calculate the new namevalue location
correctly.

The following script can trigger a kernel panic easily.

echo 'y'|mkfs.ocfs2 --fs-features=local,xattr -b 4K $DEVICE
mount -t ocfs2 $DEVICE $MNT_DIR
FILE=$MNT_DIR/$RANDOM
for((i=0;i<76;i++))
do
string_76="a$string_76"
done
string_78="aa$string_76"
string_82="aaaa$string_78"

touch $FILE
setfattr -n 'user.test1234567890' -v $string_76 $FILE
setfattr -n 'user.test1234567890' -v $string_78 $FILE
setfattr -n 'user.test1234567890' -v $string_82 $FILE

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Always try for maximum bits with new local alloc windows
Mark Fasheh [Fri, 12 Mar 2010 02:43:46 +0000 (18:43 -0800)]
ocfs2: Always try for maximum bits with new local alloc windows

What we were doing before was to ask for the current window size as the
maximum allocation. This had the effect of limiting the amount of allocation
we could get for the local alloc during times when the window size was
shrunk due to fragmentation. In some cases, that could actually *increase*
fragmentation by artificially limiting the number of bits we can accept. So
while we still want to ask for a minimum number of bits equal to window
size, there is no reason why we should limit the number of bits the local
alloc should accept. Hence always allow the maximum number of local alloc
bits.

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: set i_mode on disk during acl operations
Mark Fasheh [Mon, 15 Mar 2010 22:39:00 +0000 (15:39 -0700)]
ocfs2: set i_mode on disk during acl operations

ocfs2_set_acl() and ocfs2_init_acl() were setting i_mode on the in-memory
inode, but never setting it on the disk copy. Thus, acls were some times not
getting propagated between nodes. This patch fixes the issue by adding a
helper function ocfs2_acl_set_mode() which does this the right way.
ocfs2_set_acl() and ocfs2_init_acl() are then updated to call
ocfs2_acl_set_mode().

Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Update i_blocks in reflink operations.
Tao Ma [Wed, 10 Mar 2010 01:56:52 +0000 (09:56 +0800)]
ocfs2: Update i_blocks in reflink operations.

In reflink, we need to upate i_blocks for the target inode.

Reported-by: Jie Liu <jeff.liu@oracle.com>
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoocfs2: Change bg_chain check for ocfs2_validate_gd_parent.
Tao Ma [Wed, 3 Mar 2010 03:26:27 +0000 (11:26 +0800)]
ocfs2: Change bg_chain check for ocfs2_validate_gd_parent.

In ocfs2_validate_gd_parent, we check bg_chain against the
cl_next_free_rec of the dinode. Actually in resize, we have
the chance of bg_chain == cl_next_free_rec. So add some
additional condition check for it.

I also rename paramter "clean_error" to "resize", since the
old one is not clearly enough to indicate that we should only
meet with this case in resize.

btw, the correpsonding bug is
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1230.

Signed-off-by: Tao Ma <tao.ma@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years ago[PATCH] Skip check for mandatory locks when unlocking
Sachin Prabhu [Wed, 10 Mar 2010 15:28:40 +0000 (10:28 -0500)]
[PATCH] Skip check for mandatory locks when unlocking

ocfs2_lock() will skip locks on file which has mode set to 02666. This
is a problem in cases where the mode of the file is changed after a
process has obtained a lock on the file.

ocfs2_lock() should skip the check for mandatory locks when unlocking a
file.

Signed-off-by: Sachin Prabhu <sprabhu@redhat.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
14 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux...
Linus Torvalds [Mon, 15 Mar 2010 03:29:21 +0000 (20:29 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/lenb/linux-acpi-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux-acpi-2.6: (34 commits)
  ACPI: processor: push file static MADT pointer into internal map_madt_entry()
  ACPI: processor: refactor internal map_lsapic_id()
  ACPI: processor: refactor internal map_x2apic_id()
  ACPI: processor: refactor internal map_lapic_id()
  ACPI: processor: driver doesn't need to evaluate _PDC
  ACPI: processor: remove early _PDC optin quirks
  ACPI: processor: add internal processor_physically_present()
  ACPI: processor: move acpi_get_cpuid into processor_core.c
  ACPI: processor: export acpi_get_cpuid()
  ACPI: processor: mv processor_pdc.c processor_core.c
  ACPI: processor: mv processor_core.c processor_driver.c
  ACPI: plan to delete "acpi=ht" boot option
  ACPI: remove "acpi=ht" DMI blacklist
  PNPACPI: add bus number support
  PNPACPI: add window support
  resource: add window support
  resource: add bus number support
  resource: expand IORESOURCE_TYPE_BITS to make room for bus resource type
  acpiphp: Execute ACPI _REG method for hotadded devices
  ACPI video: Be more liberal in validating _BQC behaviour
  ...

14 years agoinit dynamic bin_attribute structures
Wolfram Sang [Mon, 15 Mar 2010 00:29:41 +0000 (01:29 +0100)]
init dynamic bin_attribute structures

Commit 6992f5334995af474c2b58d010d08bc597f0f2fe ("sysfs: Use one lockdep
class per sysfs attribute.") introduced this requirement.  First, at25
was fixed manually.  Then, other occurences were found with coccinelle
and the following semantic patch.  Results were reviewed and fixed up:

    @ init @
    identifier struct_name, bin;
    @@

     struct struct_name {
     ...
     struct bin_attribute bin;
     ...
     };

    @ main extends init @
    expression E;
    statement S;
    identifier name, err;
    @@

    (
     struct struct_name *name;
    |
    - struct struct_name *name = NULL;
    + struct struct_name *name;
    )
     ...
    (
     sysfs_bin_attr_init(&name->bin);
    |
    + sysfs_bin_attr_init(&name->bin);
     if (sysfs_create_bin_file(E, &name->bin))
     S
    |
    + sysfs_bin_attr_init(&name->bin);
     err = sysfs_create_bin_file(E, &name->bin);
    )

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branches 'battery-2.6.34', 'bugzilla-10805', 'bugzilla-14668', 'bugzilla-531916...
Len Brown [Mon, 15 Mar 2010 01:30:17 +0000 (21:30 -0400)]
Merge branches 'battery-2.6.34', 'bugzilla-10805', 'bugzilla-14668', 'bugzilla-531916-power-state', 'ht-warn-2.6.34', 'pnp', 'processor-rename', 'sony-2.6.34', 'suse-bugzilla-531547', 'tz-check', 'video' and 'misc-2.6.34' into release

14 years agoACPI: processor: push file static MADT pointer into internal map_madt_entry()
Alex Chiang [Mon, 22 Feb 2010 19:12:00 +0000 (12:12 -0700)]
ACPI: processor: push file static MADT pointer into internal map_madt_entry()

There's no real need for a pointer to the MADT to be global. The only
function who uses it is map_madt_entry.

This allows us to remove some more ugly #ifdefs.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: refactor internal map_lsapic_id()
Alex Chiang [Mon, 22 Feb 2010 19:11:55 +0000 (12:11 -0700)]
ACPI: processor: refactor internal map_lsapic_id()

Un-nest the if statements for readability.

Remove comments that re-state the obvious.

Change the control flow so that we no longer need a temp variable.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: refactor internal map_x2apic_id()
Alex Chiang [Mon, 22 Feb 2010 19:11:50 +0000 (12:11 -0700)]
ACPI: processor: refactor internal map_x2apic_id()

Untangle the nested if conditions to make this function look
more similar to the other map_*apic_id() functions.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: refactor internal map_lapic_id()
Alex Chiang [Mon, 22 Feb 2010 19:11:44 +0000 (12:11 -0700)]
ACPI: processor: refactor internal map_lapic_id()

Untangle the if() statement a little for readability.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: driver doesn't need to evaluate _PDC
Alex Chiang [Mon, 22 Feb 2010 19:11:39 +0000 (12:11 -0700)]
ACPI: processor: driver doesn't need to evaluate _PDC

Now that the early _PDC evaluation path knows how to correctly
evaluate _PDC on only physically present processors, there's no
need for the processor driver to evaluate it later when it loads.

To cover the hotplug case, push _PDC evaluation down into the
hotplug paths.

Cc: x86@kernel.org
Cc: Tony Luck <tony.luck@intel.com>
Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: remove early _PDC optin quirks
Alex Chiang [Mon, 22 Feb 2010 19:11:34 +0000 (12:11 -0700)]
ACPI: processor: remove early _PDC optin quirks

Now that we check for physically present processors before blindly
evaluating _PDC, we no longer need to maintain a DMI opt-in table
nor a kernel param.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: add internal processor_physically_present()
Alex Chiang [Mon, 22 Feb 2010 19:11:29 +0000 (12:11 -0700)]
ACPI: processor: add internal processor_physically_present()

Detect if a processor is physically present before evaluating _PDC.

We want this because some BIOS will provide a _PDC even for processors
that are not present. These bogus _PDC methods then attempt to load
non-existent tables, which causes problems.

Avoid those bogus landmines.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: move acpi_get_cpuid into processor_core.c
Alex Chiang [Mon, 22 Feb 2010 19:11:24 +0000 (12:11 -0700)]
ACPI: processor: move acpi_get_cpuid into processor_core.c

Enumerating processors (via MADT/_MAT) belongs in the processor core,
which is always built-in, rather than living in the processor driver
which may not be built.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: export acpi_get_cpuid()
Alex Chiang [Mon, 22 Feb 2010 19:11:19 +0000 (12:11 -0700)]
ACPI: processor: export acpi_get_cpuid()

Rename static get_cpu_id() to acpi_get_cpuid() and export it.

This change also gives us an opportunity to remove the
#ifndef CONFIG_SMP from processor_driver.c and into a header file
where it properly belongs.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: mv processor_pdc.c processor_core.c
Alex Chiang [Mon, 22 Feb 2010 19:11:14 +0000 (12:11 -0700)]
ACPI: processor: mv processor_pdc.c processor_core.c

We've renamed the old processor_core.c to processor_driver.c, to
convey the idea that it can be built modular and has driver-like
bits.

Now let's re-create a processor_core.c for the bits needed
statically by the rest of the kernel. The contents of processor_pdc.c
are a good starting spot, so let's just rename that file and
complete our three card monte.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: processor: mv processor_core.c processor_driver.c
Alex Chiang [Mon, 22 Feb 2010 19:11:08 +0000 (12:11 -0700)]
ACPI: processor: mv processor_core.c processor_driver.c

The ACPI processor driver can be built as a module. But it has
pieces of code that should always be built statically into the
kernel.

The plan is for processor_core.c to contain the static bits while
processor_driver.c contains the module-like bits.

Since the bulk of the code in the current processor_core.c is
module-like, first step is to rename the file to processor_driver.c

Next step will re-create processor_core.c and cherry-pick out
the static bits.

Acked-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Alex Chiang <achiang@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: plan to delete "acpi=ht" boot option
Len Brown [Sun, 14 Mar 2010 20:28:46 +0000 (16:28 -0400)]
ACPI: plan to delete "acpi=ht" boot option

Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI: remove "acpi=ht" DMI blacklist
Len Brown [Fri, 19 Feb 2010 04:42:47 +0000 (23:42 -0500)]
ACPI: remove "acpi=ht" DMI blacklist

SuSE added these entries when deploying ACPI in Linux-2.4.
I pulled them into Linux-2.6 on 2003-08-09.
Over the last 6+ years, several entries have proven to be
unnecessary and deleted, while no new entries have been added.
Matthew suggests that they now have negative value, and I agree.

Based-on-patch-by: Matthew Garrett <mjg59@srcf.ucam.org>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoPNPACPI: add bus number support
Bjorn Helgaas [Fri, 5 Mar 2010 17:47:57 +0000 (10:47 -0700)]
PNPACPI: add bus number support

Add support for bus number resources.  This is for bridges with a range of
bus numbers behind them.  Previously, PNP ignored bus number resources.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoPNPACPI: add window support
Bjorn Helgaas [Fri, 5 Mar 2010 17:47:52 +0000 (10:47 -0700)]
PNPACPI: add window support

Add support for resource windows.  This is for bridge resources, i.e.,
regions where a bridge forwards transactions from the primary to the
secondary side.  This does not add support for *setting* windows via
the /proc interface.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoresource: add window support
Bjorn Helgaas [Fri, 5 Mar 2010 17:47:47 +0000 (10:47 -0700)]
resource: add window support

Add support for resource windows.  This is for bridge resources, i.e.,
regions where a bridge forwards transactions from the primary to the
secondary side.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoresource: add bus number support
Bjorn Helgaas [Fri, 5 Mar 2010 17:47:42 +0000 (10:47 -0700)]
resource: add bus number support

Add support for bus number resources.  This is for bridges with a range of
bus numbers behind them.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoresource: expand IORESOURCE_TYPE_BITS to make room for bus resource type
Bjorn Helgaas [Fri, 5 Mar 2010 17:47:26 +0000 (10:47 -0700)]
resource: expand IORESOURCE_TYPE_BITS to make room for bus resource type

No functional change; this just makes room for another resource type.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agotomoyo: fix potential use after free
Dan Carpenter [Sat, 13 Mar 2010 11:14:22 +0000 (14:14 +0300)]
tomoyo: fix potential use after free

The original code returns a freed pointer.  This function is expected to
return NULL on errors.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: James Morris <jmorris@namei.org>
14 years agoacpiphp: Execute ACPI _REG method for hotadded devices
Shaohua Li [Thu, 25 Feb 2010 02:59:34 +0000 (10:59 +0800)]
acpiphp: Execute ACPI _REG method for hotadded devices

Per ACPI spec, _ERG method should be executed before device driver
gets control for hotpluged device. Firmware might do some configuration
there. See http://bugzilla.kernel.org/show_bug.cgi?id=10805. In this
machine, _REG method of docked device will configure cardbus bridge.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Tested-by: Paul Martin <pm@debian.org>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoACPI video: Be more liberal in validating _BQC behaviour
Matthew Garrett [Tue, 16 Feb 2010 21:53:50 +0000 (16:53 -0500)]
ACPI video: Be more liberal in validating _BQC behaviour

Right now, if _BQC returns a value we don't understand we immediately
invalidate it. Change this behaviour so we only invalidate it if it
continues to give an invalid answer after we've already set a brightness.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Acked-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
14 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 14 Mar 2010 18:15:45 +0000 (11:15 -0700)]
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86/mce: Fix build bug with CONFIG_PROVE_LOCKING=y && CONFIG_X86_MCE_INTEL=y

14 years agoAdd include to i2c-xii.c to fix build error
Randy Dunlap [Thu, 4 Feb 2010 20:11:09 +0000 (12:11 -0800)]
Add include to i2c-xii.c to fix build error

drivers/i2c/busses/i2c-xiic.c:493: error: implicit declaration of function 'mdelay'

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: "Richard Röjfors" <richard.rojfors@pelagicore.com>
Cc: "Ben Dooks (embedded platforms)" <ben-linux@fluff.org>
Cc: linux-i2c@vger.kernel.org
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoAdd the platform data include for the Xilinx XPS IIC Bus Interface
Richard Röjfors [Tue, 9 Mar 2010 08:17:36 +0000 (09:17 +0100)]
Add the platform data include for the Xilinx XPS IIC Bus Interface

This file was missed in the original patch that went into Linus' tree.

Cc: "Ben Dooks (embedded platforms)" <ben-linux@fluff.org>
Cc: linux-i2c@vger.kernel.org
Signed-off-by: Richard Röjfors <richard.rojfors@pelagicore.com>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sun, 14 Mar 2010 18:13:54 +0000 (11:13 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/dtor/input

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: i8042 - add ALDI/MEDION netbook E1222 to qurik reset table
  Input: ALPS - fix stuck buttons on some touchpads
  Input: wm831x-on - convert to use genirq
  Input: ads7846 - add wakeup support
  Input: appletouch - fix integer overflow issue
  Input: ad7877 - increase pen up imeout
  Input: ads7846 - add support for AD7843 parts
  Input: bf54x-keys - fix system hang when pressing a key
  Input: alps - add support for the touchpad on Toshiba Tecra A11-11L
  Input: remove BKL, fix input_open_file() locking
  Input: serio_raw - remove BKL
  Input: mousedev - remove BKL
  Input: add driver for TWL4030 vibrator device
  Input: enable remote wakeup for PNP i8042 keyboard ports
  Input: scancode in get/set_keycodes should be unsigned
  Input: i8042 - use platfrom_create_bundle() helper
  Input: wacom - merge out and in prox events
  Input: gamecon - fix off by one range check
  Input: wacom - replace WACOM_PKGLEN_PENABLED

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke...
Linus Torvalds [Sun, 14 Mar 2010 18:13:24 +0000 (11:13 -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:
  nilfs2: remove whitespaces before quoted newlines
  nilfs2: remove spaces before tabs
  nilfs2: fix various typos in comments
  nilfs2: fix typo "cout" -> "count" in error message
  nilfs2: fix function name typos in docbook comments
  nilfs2: fix discrepancy in use of static specifier

14 years agoMerge branch 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvar...
Linus Torvalds [Sun, 14 Mar 2010 18:13:04 +0000 (11:13 -0700)]
Merge branch 'i2c-for-linus' of git://git./linux/kernel/git/jdelvare/staging

* 'i2c-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  i2c-algo-bit: Add pre- and post-xfer hooks
  at24: Init dynamic bin_attribute structures
  i2c: Drop configure option I2C_DEBUG_CHIP
  tsl2550: Move from i2c/chips to misc
  i2c-i801: Don't use the block buffer for I2C block writes
  i2c-powermac: Be less verbose in the absence of real errors.
  i2c-smbus: Use device_lock/device_unlock

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh...
Linus Torvalds [Sun, 14 Mar 2010 18:11:08 +0000 (11:11 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/ericvh/v9fs

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  9p: Skip check for mandatory locks when unlocking
  9p: Fixes a simple bug enabling writes beyond 2GB.
  9p: Change the name of new protocol from 9p2010.L to 9p2000.L
  fs/9p: re-init the wstat in readdir loop
  net/9p: Add sysfs mount_tag file for virtio 9P device
  net/9p: Use the tag name in the config space for identifying mount point

14 years agox86/mce: Fix build bug with CONFIG_PROVE_LOCKING=y && CONFIG_X86_MCE_INTEL=y
Ingo Molnar [Sun, 14 Mar 2010 07:57:03 +0000 (08:57 +0100)]
x86/mce: Fix build bug with CONFIG_PROVE_LOCKING=y && CONFIG_X86_MCE_INTEL=y

Commit f56e8a076 "x86/mce: Fix RCU lockdep splats" introduced the
following build bug:

  arch/x86/kernel/cpu/mcheck/mce.c: In function 'mce_log':
  arch/x86/kernel/cpu/mcheck/mce.c:166: error: 'mce_read_mutex' undeclared (first use in this function)
  arch/x86/kernel/cpu/mcheck/mce.c:166: error: (Each undeclared identifier is reported only once
  arch/x86/kernel/cpu/mcheck/mce.c:166: error: for each function it appears in.)

Move the in-the-middle-of-file lock variable up to the variable
definition section, the top of the .c file.

Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: x86@kernel.org
Cc: laijs@cn.fujitsu.com
Cc: dipankar@in.ibm.com
Cc: mathieu.desnoyers@polymtl.ca
Cc: josh@joshtriplett.org
Cc: dvhltc@us.ibm.com
Cc: niv@us.ibm.com
Cc: peterz@infradead.org
Cc: rostedt@goodmis.org
Cc: Valdis.Kletnieks@vt.edu
Cc: dhowells@redhat.com
LKML-Reference: <1267830207-9474-3-git-send-email-paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoInput: i8042 - add ALDI/MEDION netbook E1222 to qurik reset table
Christoph Fritz [Sun, 14 Mar 2010 06:26:23 +0000 (22:26 -0800)]
Input: i8042 - add ALDI/MEDION netbook E1222 to qurik reset table

ALDI/MEDION netbook E1222 needs to be in the reset quirk list for
its touchpad's proper function.

Reported-by: Michael Fischer <mifi@gmx.de>
Signed-off-by: Christoph Fritz <chf.fritz@googlemail.com>
Cc: stable@kernel.org
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
14 years agoInput: ALPS - fix stuck buttons on some touchpads
Martin Buck [Sun, 14 Mar 2010 06:23:58 +0000 (22:23 -0800)]
Input: ALPS - fix stuck buttons on some touchpads

Enable button release event redirection to the device that got the
button press not only for touchpads with interleaved protocols, but
unconditionally for all Alps touchpads. This is required at least
for the touchpads in Dell Inspiron 8200 and Latitude d630.

Signed-off-by: Martin Buck <mb-tmp-yvahk-vachg@gromit.dyndns.org>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Sun, 14 Mar 2010 05:29:38 +0000 (21:29 -0800)]
Merge git://git./linux/kernel/git/davem/sparc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc64: Add very basic XVR-1000 framebuffer driver.
  qlogicpti: Remove slash in QlogicPTI irq name

14 years agonilfs2: remove whitespaces before quoted newlines
Ryusuke Konishi [Sat, 13 Mar 2010 19:01:27 +0000 (04:01 +0900)]
nilfs2: remove whitespaces before quoted newlines

This kills the following checkpatch warnings:

 WARNING: unnecessary whitespace before a quoted newline
 #869: FILE: super.c:869:
 +                 "remount to a different snapshot. \n",

 WARNING: unnecessary whitespace before a quoted newline
 #389: FILE: the_nilfs.c:389:
 +          printk(KERN_ERR "NILFS: too short segment. \n");

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: remove spaces before tabs
Ryusuke Konishi [Sat, 13 Mar 2010 18:55:56 +0000 (03:55 +0900)]
nilfs2: remove spaces before tabs

This kills the following checkpatch warnings:

 WARNING: please, no space before tabs
 #74: FILE: segment.h:74:
 +^Iunsigned ^I^Iflags;$

 WARNING: please, no space before tabs
 #35: FILE: segbuf.c:35:
 +^Iint ^I^I^Istart, end; /* The region to be submitted */$

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: fix various typos in comments
Ryusuke Konishi [Sat, 13 Mar 2010 18:32:40 +0000 (03:32 +0900)]
nilfs2: fix various typos in comments

This fixes various typos I found in comments of nilfs2.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: fix typo "cout" -> "count" in error message
Ryusuke Konishi [Sat, 13 Mar 2010 18:17:45 +0000 (03:17 +0900)]
nilfs2: fix typo "cout" -> "count" in error message

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: fix function name typos in docbook comments
Ryusuke Konishi [Sat, 13 Mar 2010 18:01:03 +0000 (03:01 +0900)]
nilfs2: fix function name typos in docbook comments

Fixes the following typos in docbook comments:

 nilfs_detroy_transaction_cache -> nilfs_destroy_transaction_cache
 nilfs_secgtor_start_timer -> nilfs_segctor_start_timer

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agonilfs2: fix discrepancy in use of static specifier
Ryusuke Konishi [Sat, 13 Mar 2010 17:01:51 +0000 (02:01 +0900)]
nilfs2: fix discrepancy in use of static specifier

Two segbuf functions, nilfs_segbuf_write and nilfs_segbuf_wait, are
declared with the static storage class specifier, but their
implementations are not.

This fixes the discrepancy.

Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
14 years agosparc64: Add very basic XVR-1000 framebuffer driver.
David S. Miller [Sun, 14 Mar 2010 00:25:03 +0000 (16:25 -0800)]
sparc64: Add very basic XVR-1000 framebuffer driver.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Frans van Berckel <fberckel@xs4all.nl>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 13 Mar 2010 22:50:18 +0000 (14:50 -0800)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (108 commits)
  bridge: ensure to unlock in error path in br_multicast_query().
  drivers/net/tulip/eeprom.c: fix bogus "(null)" in tulip init messages
  sky2: Avoid rtnl_unlock without rtnl_lock
  ipv6: Send netlink notification when DAD fails
  drivers/net/tg3.c: change the field used with the TG3_FLAG_10_100_ONLY constant
  ipconfig: Handle devices which take some time to come up.
  mac80211: Fix memory leak in ieee80211_if_write()
  mac80211: Fix (dynamic) power save entry
  ipw2200: use kmalloc for large local variables
  ath5k: read eeprom IQ calibration values correctly for G mode
  ath5k: fix I/Q calibration (for real)
  ath5k: fix TSF reset
  ath5k: use fixed antenna for tx descriptors
  libipw: split ieee->networks into small pieces
  mac80211: Fix sta_mtx unlocking on insert STA failure path
  rt2x00: remove KSEG1ADDR define from rt2x00soc.h
  net: add ColdFire support to the smc91x driver
  asix: fix setting mac address for AX88772
  ipv6 ip6_tunnel: eliminate unused recursion field from ip6_tnl{}.
  net: Fix dev_mc_add()
  ...

14 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 13 Mar 2010 22:46:18 +0000 (14:46 -0800)]
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: Fix pick_next_highest_task_rt() for cgroups
  sched: Cleanup: remove unused variable in try_to_wake_up()
  x86: Fix sched_clock_cpu for systems with unsynchronized TSC

14 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 13 Mar 2010 22:45:49 +0000 (14:45 -0800)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86, k8 nb: Fix boot crash: enable k8_northbridges unconditionally on AMD systems
  x86, UV: Fix target_cpus() in x2apic_uv_x.c
  x86: Reduce per cpu warning boot up messages
  x86: Reduce per cpu MCA boot up messages
  x86_64, cpa: Don't work hard in preserving kernel 2M mappings when using 4K already

14 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 13 Mar 2010 22:43:01 +0000 (14:43 -0800)]
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  locking: Make sparse work with inline spinlocks and rwlocks
  x86/mce: Fix RCU lockdep splats
  rcu: Increase RCU CPU stall timeouts if PROVE_RCU
  ftrace: Replace read_barrier_depends() with rcu_dereference_raw()
  rcu: Suppress RCU lockdep warnings during early boot
  rcu, ftrace: Fix RCU lockdep splat in ftrace_perf_buf_prepare()
  rcu: Suppress __mpol_dup() false positive from RCU lockdep
  rcu: Make rcu_read_lock_sched_held() handle !PREEMPT
  rcu: Add control variables to lockdep_rcu_dereference() diagnostics
  rcu, cgroup: Relax the check in task_subsys_state() as early boot is now handled by lockdep-RCU
  rcu: Use wrapper function instead of exporting tasklist_lock
  sched, rcu: Fix rcu_dereference() for RCU-lockdep
  rcu: Make task_subsys_state() RCU-lockdep checks handle boot-time use
  rcu: Fix holdoff for accelerated GPs for last non-dynticked CPU
  x86/gart: Unexport gart_iommu_aperture

Fix trivial conflicts in kernel/trace/ftrace.c

14 years agoMerge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 13 Mar 2010 22:40:50 +0000 (14:40 -0800)]
Merge branch 'tracing-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  tracing: Do not record user stack trace from NMI context
  tracing: Disable buffer switching when starting or stopping trace
  tracing: Use same local variable when resetting the ring buffer
  function-graph: Init curr_ret_stack with ret_stack
  ring-buffer: Move disabled check into preempt disable section
  function-graph: Add tracing_thresh support to function_graph tracer
  tracing: Update the comm field in the right variable in update_max_tr
  function-graph: Use comment notation for func names of dangling '}'
  function-graph: Fix unused reference to ftrace_set_func()
  tracing: Fix warning in s_next of trace file ops
  tracing: Include irqflags headers from trace clock

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Sat, 13 Mar 2010 22:39:54 +0000 (14:39 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda - Sound MSI fallout on a Asus mobo NVIDIA MCP55
  sound: fix opti92x-ad1848 build
  ALSA: hda - Fix input source elements of secondary ADCs on Realtek
  ALSA: hda - Fix wrong model range check for ALC268

14 years agoMerge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 13 Mar 2010 22:39:42 +0000 (14:39 -0800)]
Merge branch 'perf-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  perf: Provide generic perf_sample_data initialization
  MAINTAINERS: Add Arnaldo as tools/perf/ co-maintainer
  perf trace: Don't use pager if scripting
  perf trace/scripting: Remove extraneous header read
  perf, ARM: Modify kuser rmb() call to compile for Thumb-2
  x86/stacktrace: Don't dereference bad frame pointers
  perf archive: Don't try to collect files without a build-id
  perf_events, x86: Fixup fixed counter constraints
  perf, x86: Restrict the ANY flag
  perf, x86: rename macro in ARCH_PERFMON_EVENTSEL_ENABLE
  perf, x86: add some IBS macros to perf_event.h
  perf, x86: make IBS macros available in perf_event.h
  hw-breakpoints: Remove stub unthrottle callback
  x86/hw-breakpoints: Remove the name field
  perf: Remove pointless breakpoint union
  perf lock: Drop the buffers multiplexing dependency
  perf lock: Fix and add misc documentally things
  percpu: Add __percpu sparse annotations to hw_breakpoint

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes
Linus Torvalds [Sat, 13 Mar 2010 22:38:53 +0000 (14:38 -0800)]
Merge git://git./linux/kernel/git/steve/gfs2-2.6-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6-fixes:
  GFS2: Skip check for mandatory locks when unlocking
  GFS2: Allow the number of committed revokes to temporarily be negative
  GFS2: do not select QUOTA

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Sat, 13 Mar 2010 22:38:31 +0000 (14:38 -0800)]
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:
  RDMA/nes: Fix CX4 link problem in back-to-back configuration
  RDMA/nes: Clear stall bit before destroying NIC QP
  RDMA/nes: Set assume_aligned_header bit
  RDMA/cxgb3: Wait at least one schedule cycle during device removal
  IB/mad: Ignore iWARP devices on device removal
  IPoIB: Include return code in trace message for ib_post_send() failures
  IPoIB: Fix TX queue lockup with mixed UD/CM traffic

14 years agobridge: ensure to unlock in error path in br_multicast_query().
YOSHIFUJI Hideaki [Sat, 13 Mar 2010 20:27:21 +0000 (12:27 -0800)]
bridge: ensure to unlock in error path in br_multicast_query().

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/tulip/eeprom.c: fix bogus "(null)" in tulip init messages
Joe Perches [Sat, 13 Mar 2010 20:26:15 +0000 (12:26 -0800)]
drivers/net/tulip/eeprom.c: fix bogus "(null)" in tulip init messages

On Wed, 2010-03-10 at 08:41 -0800, David Miller wrote:
> From: Mikael Pettersson <mikpe@it.uu.se>
> Date: Wed, 10 Mar 2010 16:33:28 +0100
> > Booting 2.6.34-rc1 on a machine with a tulip nic I see
> > a number of kernel messages that include "(null)" where
> > previous kernels included the string "tulip0":
> CC:'ing the guilty party :-)  It's one of the following
> commits:

Thanks Mikael.

Anonymity has some good attributes.
Blame avoidance is one of them.

I've broad shoulders.  It's me, then Dwight Howard...

There might be another few of these where ->name or ->dev
was used before struct device or net_device was registered.
I'll go back and check.

tulip_core has:

if (tp->flags & HAS_MEDIA_TABLE) {
sprintf(dev->name, DRV_NAME "%d", board_idx); /* hack */
tulip_parse_eeprom(dev);
strcpy(dev->name, "eth%d"); /* un-hack */
}

So I don't feel _too_ bad.

tulip_parse_eeprom is done before register_netdev so the logging
there can not use netdev_<level> or dev_<level>(&dev->dev

Signed-off-by: Joe Perches <joe@perches.com>
Tested-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agosky2: Avoid rtnl_unlock without rtnl_lock
Mike McCormack [Sat, 13 Mar 2010 20:24:18 +0000 (12:24 -0800)]
sky2: Avoid rtnl_unlock without rtnl_lock

Make sure we always call rtnl_lock before going down the
error path in sky2_resume, which unlocks the rtnl lock.

Signed-off-by: Mike McCormack <mikem@ring3k.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoipv6: Send netlink notification when DAD fails
Herbert Xu [Sat, 13 Mar 2010 20:23:29 +0000 (12:23 -0800)]
ipv6: Send netlink notification when DAD fails

If we are managing IPv6 addresses using DHCP, it would be nice
for user-space to be notified if an address configured through
DHCP fails DAD.  Otherwise user-space would have to poll to see
whether DAD succeeds.

This patch uses the existing notification mechanism and simply
hooks it into the DAD failure code path.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agodrivers/net/tg3.c: change the field used with the TG3_FLAG_10_100_ONLY constant
Julia Lawall [Sat, 13 Mar 2010 20:22:16 +0000 (12:22 -0800)]
drivers/net/tg3.c: change the field used with the TG3_FLAG_10_100_ONLY constant

The constant TG3_FLAG_10_100_ONLY should be used with the tg3_flags field,
not the tg3_flags2 field, as done elsewhere in the same file.

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>
14 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Sat, 13 Mar 2010 20:17:09 +0000 (12:17 -0800)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

14 years agoi2c-algo-bit: Add pre- and post-xfer hooks
Jean Delvare [Sat, 13 Mar 2010 19:56:56 +0000 (20:56 +0100)]
i2c-algo-bit: Add pre- and post-xfer hooks

Drivers might have to do random things before and/or after I2C
transfers. Add hooks to the i2c-algo-bit implementation to let them do
so.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Alex Deucher <alexdeucher@gmail.com>
14 years agoat24: Init dynamic bin_attribute structures
Wolfram Sang [Sat, 13 Mar 2010 19:56:55 +0000 (20:56 +0100)]
at24: Init dynamic bin_attribute structures

Commit 6992f5334995af474c2b58d010d08bc597f0f2fe introduced this requirement.

Reported-by: Albrecht Dress <albrecht.dress@arcor.de>
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
14 years agoi2c: Drop configure option I2C_DEBUG_CHIP
Jean Delvare [Sat, 13 Mar 2010 19:56:55 +0000 (20:56 +0100)]
i2c: Drop configure option I2C_DEBUG_CHIP

Now that directory drivers/i2c/chips is gone, configuration option
I2C_DEBUG_CHIP no longer has any effect, so we can drop it.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
14 years agotsl2550: Move from i2c/chips to misc
Jean Delvare [Sat, 13 Mar 2010 19:56:54 +0000 (20:56 +0100)]
tsl2550: Move from i2c/chips to misc

Move the last remaining driver from i2c/chips to misc. Good ridance!

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Wolfram Sang <w.sang@pengutronix.de>
Acked-by: Jonathan Cameron <jic23@cam.ac.uk>
14 years agoi2c-i801: Don't use the block buffer for I2C block writes
Jean Delvare [Sat, 13 Mar 2010 19:56:53 +0000 (20:56 +0100)]
i2c-i801: Don't use the block buffer for I2C block writes

Experience has shown that the block buffer can only be used for SMBus
(not I2C) block transactions, even though the datasheet doesn't
mention this limitation.

Reported-by: Felix Rubinstein <felixru@gmail.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Oleg Ryjkov <oryjkov@gmail.com>
Cc: stable@kernel.org
14 years agoi2c-powermac: Be less verbose in the absence of real errors.
Jean Delvare [Sat, 13 Mar 2010 19:56:52 +0000 (20:56 +0100)]
i2c-powermac: Be less verbose in the absence of real errors.

Be less verbose in the absence of real errors. We don't have to report
failed probes to the users, it's only confusing them.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Tested-by: Andrey Gusev <ronne@list.ru>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: stable@kernel.org
14 years agoi2c-smbus: Use device_lock/device_unlock
Stephen Rothwell [Mon, 1 Mar 2010 05:04:45 +0000 (16:04 +1100)]
i2c-smbus: Use device_lock/device_unlock

Use the new device locking/unlocking API.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
14 years ago9p: Skip check for mandatory locks when unlocking
Sachin Prabhu [Sat, 13 Mar 2010 15:03:55 +0000 (09:03 -0600)]
9p: Skip check for mandatory locks when unlocking

While investigating a bug, I came across a possible bug in v9fs. The
problem is similar to the one reported for NFS by ASANO Masahiro in
http://lkml.org/lkml/2005/12/21/334.

v9fs_file_lock() will skip locks on file which has mode set to 02666.
This is a problem in cases where the mode of the file is changed after
a process has obtained a lock on the file. Such a lock will be skipped
during unlock and the machine will end up with a BUG in
locks_remove_flock().

v9fs_file_lock() should skip the check for mandatory locks when
unlocking a file.

Signed-off-by: Sachin Prabhu <sprabhu@redhat.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
14 years ago9p: Fixes a simple bug enabling writes beyond 2GB.
jvrao [Mon, 8 Mar 2010 22:07:02 +0000 (22:07 +0000)]
9p: Fixes a simple bug enabling writes beyond 2GB.

Fixes a simple bug so that large files beyond 2GB can be created.

Signed-off-by: Venkateswararao Jujjuri <jvrao@linux.vnet.ibm.com>
Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
14 years ago9p: Change the name of new protocol from 9p2010.L to 9p2000.L
Sripathi Kodi [Mon, 8 Mar 2010 17:33:04 +0000 (17:33 +0000)]
9p: Change the name of new protocol from 9p2010.L to 9p2000.L

This patch changes the name of the new 9P protocol from 9p2010.L to
9p2000.u. This is because we learnt that the name 9p2010 is already
being used by others.

Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
14 years agofs/9p: re-init the wstat in readdir loop
Aneesh Kumar K.V [Sat, 6 Mar 2010 04:44:16 +0000 (04:44 +0000)]
fs/9p: re-init the wstat in readdir loop

This ensure that on failure when we free the stat buf we don't end up
freeing an already freed pointer in the earlier loop

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Sripathi Kodi <sripathik@in.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
14 years agonet/9p: Add sysfs mount_tag file for virtio 9P device
Aneesh Kumar K.V [Sat, 6 Mar 2010 04:44:15 +0000 (04:44 +0000)]
net/9p: Add sysfs mount_tag file for virtio 9P device

This adds a new file for virtio 9P device. The file
contain details of the mount device name that should
be used to mount the 9P file system.

Ex: /sys/devices/virtio-pci/virtio1/mount_tag  file now
contian the tag name to be used to mount the 9P file system.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
14 years agonet/9p: Use the tag name in the config space for identifying mount point
Aneesh Kumar K.V [Sat, 6 Mar 2010 04:44:14 +0000 (04:44 +0000)]
net/9p: Use the tag name in the config space for identifying mount point

This patch use the tag name in the config space to identify the
mount device. The the virtio device name depend on the enumeration
order of the device and may not remain the same across multiple boots
So we use the tag name which is set via qemu option to uniquely identify
the mount device

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
14 years agox86, k8 nb: Fix boot crash: enable k8_northbridges unconditionally on AMD systems
Borislav Petkov [Fri, 12 Mar 2010 14:43:03 +0000 (15:43 +0100)]
x86, k8 nb: Fix boot crash: enable k8_northbridges unconditionally on AMD systems

de957628ce7c84764ff41331111036b3ae5bad0f changed setting of the
x86_init.iommu.iommu_init function ptr only when GART IOMMU is
found.

One side effect of it is that num_k8_northbridges
is not initialized anymore if not explicitly
called. This resulted in uninitialized pointers in
<arch/x86/kernel/cpu/intel_cacheinfo.c:amd_calc_l3_indices()>,
for example, which uses the num_k8_northbridges thing through
node_to_k8_nb_misc().

Fix that through an initcall that runs right after the PCI
subsystem and does all the scanning. Then, remove initialization
in gart_iommu_init() which is a rootfs_initcall and we're
running before that.

What is more, since num_k8_northbridges is being used in other
places beside GART IOMMU, include it whenever we add AMD CPU
support. The previous dependency chain in kconfig contained

K8_NB depends on AGP_AMD64|GART_IOMMU

which was clearly incorrect. The more natural way in terms of
hardware dependency should be

AGP_AMD64|GART_IOMMU depends on K8_NB depends on CPU_SUP_AMD &&
PCI. Make it so Number One!

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Joerg Roedel <joerg.roedel@amd.com>
LKML-Reference: <20100312144303.GA29262@aftab>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Tested-by: Joerg Roedel <joerg.roedel@amd.com>
14 years agotracing: Do not record user stack trace from NMI context
Steven Rostedt [Sat, 13 Mar 2010 01:03:30 +0000 (20:03 -0500)]
tracing: Do not record user stack trace from NMI context

A bug was found with Li Zefan's ftrace_stress_test that caused applications
to segfault during the test.

Placing a tracing_off() in the segfault code, and examining several
traces, I found that the following was always the case. The lock tracer
was enabled (lockdep being required) and userstack was enabled. Testing
this out, I just enabled the two, but that was not good enough. I needed
to run something else that could trigger it. Running a load like hackbench
did not work, but executing a new program would. The following would
trigger the segfault within seconds:

  # echo 1 > /debug/tracing/options/userstacktrace
  # echo 1 > /debug/tracing/events/lock/enable
  # while :; do ls > /dev/null ; done

Enabling the function graph tracer and looking at what was happening
I finally noticed that all cashes happened just after an NMI.

 1)               |    copy_user_handle_tail() {
 1)               |      bad_area_nosemaphore() {
 1)               |        __bad_area_nosemaphore() {
 1)               |          no_context() {
 1)               |            fixup_exception() {
 1)   0.319 us    |              search_exception_tables();
 1)   0.873 us    |            }
[...]
 1)   0.314 us    |  __rcu_read_unlock();
 1)   0.325 us    |    native_apic_mem_write();
 1)   0.943 us    |  }
 1)   0.304 us    |  rcu_nmi_exit();
[...]
 1)   0.479 us    |  find_vma();
 1)               |  bad_area() {
 1)               |    __bad_area() {

After capturing several traces of failures, all of them happened
after an NMI. Curious about this, I added a trace_printk() to the NMI
handler to read the regs->ip to see where the NMI happened. In which I
found out it was here:

ffffffff8135b660 <page_fault>:
ffffffff8135b660:       48 83 ec 78             sub    $0x78,%rsp
ffffffff8135b664:       e8 97 01 00 00          callq  ffffffff8135b800 <error_entry>

What was happening is that the NMI would happen at the place that a page
fault occurred. It would call rcu_read_lock() which was traced by
the lock events, and the user_stack_trace would run. This would trigger
a page fault inside the NMI. I do not see where the CR2 register is
saved or restored in NMI handling. This means that it would corrupt
the page fault handling that the NMI interrupted.

The reason the while loop of ls helped trigger the bug, was that
each execution of ls would cause lots of pages to be faulted in, and
increase the chances of the race happening.

The simple solution is to not allow user stack traces in NMI context.
After this patch, I ran the above "ls" test for a couple of hours
without any issues. Without this patch, the bug would trigger in less
than a minute.

Cc: stable@kernel.org
Reported-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
14 years agotracing: Disable buffer switching when starting or stopping trace
Steven Rostedt [Sat, 13 Mar 2010 00:56:00 +0000 (19:56 -0500)]
tracing: Disable buffer switching when starting or stopping trace

When the trace iterator is read, tracing_start() and tracing_stop()
is called to stop tracing while the iterator is processing the trace
output.

These functions disable both the standard buffer and the max latency
buffer. But if the wakeup tracer is running, it can switch these
buffers between the two disables:

  buffer = global_trace.buffer;
  if (buffer)
      ring_buffer_record_disable(buffer);

      <<<--------- swap happens here

  buffer = max_tr.buffer;
  if (buffer)
      ring_buffer_record_disable(buffer);

What happens is that we disabled the same buffer twice. On tracing_start()
we can enable the same buffer twice. All ring_buffer_record_disable()
must be matched with a ring_buffer_record_enable() or the buffer
can be disable permanently, or enable prematurely, and cause a bug
where a reset happens while a trace is commiting.

This patch protects these two by taking the ftrace_max_lock to prevent
a switch from occurring.

Found with Li Zefan's ftrace_stress_test.

Cc: stable@kernel.org
Reported-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
14 years agotracing: Use same local variable when resetting the ring buffer
Steven Rostedt [Sat, 13 Mar 2010 00:48:41 +0000 (19:48 -0500)]
tracing: Use same local variable when resetting the ring buffer

In the ftrace code that resets the ring buffer it references the
buffer with a local variable, but then uses the tr->buffer as the
parameter to reset. If the wakeup tracer is running, which can
switch the tr->buffer with the max saved buffer, this can break
the requirement of disabling the buffer before the reset.

   buffer = tr->buffer;
   ring_buffer_record_disable(buffer);
   synchronize_sched();
   __tracing_reset(tr->buffer, cpu);

If the tr->buffer is swapped, then the reset is not happening to the
buffer that was disabled. This will cause the ring buffer to fail.

Found with Li Zefan's ftrace_stress_test.

Cc: stable@kernel.org
Reported-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>