GitHub/exynos8895/android_kernel_samsung_universal8895.git
15 years agomd/raid5: allow layout and chunksize to be changed on active array.
NeilBrown [Tue, 31 Mar 2009 04:24:23 +0000 (15:24 +1100)]
md/raid5: allow layout and chunksize to be changed on active array.

If an array has 3 or more devices, we allow the chunksize or layout
to be changed and when a reshape starts, we use these as the 'new'
values.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: reshape using largest of old and new chunk size
NeilBrown [Tue, 31 Mar 2009 04:21:40 +0000 (15:21 +1100)]
md/raid5: reshape using largest of old and new chunk size

This ensures that even when old and new stripes are overlapping,
we will try to read all of the old before having to write any
of the new.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: prepare for allowing reshape to change layout
NeilBrown [Tue, 31 Mar 2009 04:20:22 +0000 (15:20 +1100)]
md/raid5: prepare for allowing reshape to change layout

Add prev_algo to raid5_conf_t along the same lines as prev_chunk
and previous_raid_disks.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: prepare for allowing reshape to change chunksize.
NeilBrown [Tue, 31 Mar 2009 04:19:07 +0000 (15:19 +1100)]
md/raid5: prepare for allowing reshape to change chunksize.

Add "prev_chunk" to raid5_conf_t, similar to "previous_raid_disks", to
remember what the chunk size was before the reshape that is currently
underway.

This seems like duplication with "chunk_size" and "new_chunk" in
mddev_t, and to some extent it is, but there are differences.
The values in mddev_t are always defined and often the same.
The prev* values are only defined if a reshape is underway.

Also (and more significantly) the raid5_conf_t values will be changed
at the same time (inside an appropriate lock) that the reshape is
started by setting reshape_position.  In contrast, the new_chunk value
is set when the sysfs file is written which could be well before the
reshape starts.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: clearly differentiate 'before' and 'after' stripes during reshape.
NeilBrown [Tue, 31 Mar 2009 04:19:03 +0000 (15:19 +1100)]
md/raid5: clearly differentiate 'before' and 'after' stripes during reshape.

During a raid5 reshape, we have some stripes in the cache that are
'before' the reshape (and are still to be processed) and some that are
'after'.  They are currently differentiated by having different
->disks values as the only reshape current supported involves changing
the number of disks.

However we will soon support reshapes that do not change the number
of disks (chunk parity or chunk size).  So make the difference more
explicit with a 'generation' number.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agoDocumentation/md.txt update
NeilBrown [Tue, 31 Mar 2009 04:18:37 +0000 (15:18 +1100)]
Documentation/md.txt update

Update md.txt to reflect recent changes in a number of sysfs
attributes.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: allow number of drives in raid5 to be reduced
NeilBrown [Tue, 31 Mar 2009 04:17:38 +0000 (15:17 +1100)]
md: allow number of drives in raid5 to be reduced

When reshaping a raid5 to have fewer devices, we work from the end of
the array to the beginning.
md_do_sync gives addresses to sync_request that go from the beginning
to the end.  So largely ignore them use the internal state variable
"reshape_progress" to keep track of what to do next.

Never allow the size to be reduced below the minimum (4 for raid6,
3 otherwise).

We require that the size of the array has already been reduced before
the array is reshaped to a smaller size.  This is because simply
reducing the size is an easily reversible operation, while the reshape
is immediately destructive and so is not reversible for the blocks at
the ends of the devices.
Thus to reshape an array to have fewer devices, you must first write
an appropriately small size to md/array_size.

When reshape finished, we remove any drives that are no longer
needed and fix up ->degraded.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: change reshape-progress measurement to cope with reshaping backwards.
NeilBrown [Tue, 31 Mar 2009 04:16:46 +0000 (15:16 +1100)]
md/raid5: change reshape-progress measurement to cope with reshaping backwards.

When reducing the number of devices in a raid4/5/6, the reshape
process has to start at the end of the array and work down to the
beginning.  So we need to handle expand_progress and expand_lo
differently.

This patch renames "expand_progress" and "expand_lo" to avoid the
implication that anything is getting bigger (expand->reshape) and
every place they are used, we make sure that they are used the right
way depending on whether delta_disks is positive or negative.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: add explicit method to signal the end of a reshape.
NeilBrown [Tue, 31 Mar 2009 04:15:05 +0000 (15:15 +1100)]
md: add explicit method to signal the end of a reshape.

Currently raid5 (the only module that supports restriping)
notices that the reshape has finished be sync_request being
given a large value, and handles any cleanup them.

This patch changes it so md_check_recovery calls into an
explicit finish_reshape method as well.

The clean-up from sync_request can do things that need to be
done promptly, typically things local to the raid5_conf_t
structure.

The "finish_reshape" method is called under the mddev_lock
so it can do things involving reconfiguring the device.

This allows us to get rid of md_set_array_sectors_locked, which
would have caused a deadlock if you tried to stop and array
while a reshape was happening.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: enhance raid5_size to work correctly with negative delta_disks
NeilBrown [Tue, 31 Mar 2009 04:10:36 +0000 (15:10 +1100)]
md/raid5: enhance raid5_size to work correctly with negative delta_disks

This is the first of four patches which combine to allow md/raid5 to
reduce the number of devices in the array by restriping the data over
a subset of the devices.

If the number of disks in a raid4/5/6 is being reduced, then the
default size must be based on the new number, not the old number
of devices.
In general, it should be based on the smaller of new and old.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: drop qd_idx from r6_state
NeilBrown [Tue, 31 Mar 2009 04:10:16 +0000 (15:10 +1100)]
md/raid5: drop qd_idx from r6_state

We now have this value in stripe_head so we don't need to duplicate
it.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid6: move raid6 data processing to raid6_pq.ko
Dan Williams [Tue, 31 Mar 2009 04:09:39 +0000 (15:09 +1100)]
md/raid6: move raid6 data processing to raid6_pq.ko

Move the raid6 data processing routines into a standalone module
(raid6_pq) to prepare them to be called from async_tx wrappers and other
non-md drivers/modules.  This precludes a circular dependency of raid456
needing the async modules for data processing while those modules in
turn depend on raid456 for the base level synchronous raid6 routines.

To support this move:
1/ The exportable definitions in raid6.h move to include/linux/raid/pq.h
2/ The raid6_call, recovery calls, and table symbols are exported
3/ Extra #ifdef __KERNEL__ statements to enable the userspace raid6test to
   compile

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: raid5 run(): Fix max_degraded for raid level 4.
Andre Noll [Tue, 31 Mar 2009 04:00:56 +0000 (15:00 +1100)]
md: raid5 run(): Fix max_degraded for raid level 4.

raid4 allows only one failed disk.

Signed-off-by: Andre Noll <maan@systemlinux.org>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: 'array_size' sysfs attribute
Dan Williams [Tue, 31 Mar 2009 04:00:31 +0000 (15:00 +1100)]
md: 'array_size' sysfs attribute

Allow userspace to set the size of the array according to the following
semantics:

1/ size must be <= to the size returned by mddev->pers->size(mddev, 0, 0)
   a) If size is set before the array is running, do_md_run will fail
      if size is greater than the default size
   b) A reshape attempt that reduces the default size to less than the set
      array size should be blocked
2/ once userspace sets the size the kernel will not change it
3/ writing 'default' to this attribute returns control of the size to the
   kernel and reverts to the size reported by the personality

Also, convert locations that need to know the default size from directly
reading ->array_sectors to <pers>_size.  Resync/reshape operations
always follow the default size.

Finally, fixup other locations that read a number of 1k-blocks from
userspace to use strict_blocks_to_sectors() which checks for unsigned
long long to sector_t overflow and blocks to sectors overflow.

Reviewed-by: Andre Noll <maan@systemlinux.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
15 years agomd: centralize ->array_sectors modifications
Dan Williams [Tue, 31 Mar 2009 03:59:03 +0000 (14:59 +1100)]
md: centralize ->array_sectors modifications

Get personalities out of the business of directly modifying
->array_sectors.  Lays groundwork to introduce policy on when
->array_sectors can be modified.

Reviewed-by: Andre Noll <maan@systemlinux.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
15 years agomd: add 'size' as a personality method
Dan Williams [Wed, 18 Mar 2009 01:10:40 +0000 (18:10 -0700)]
md: add 'size' as a personality method

In preparation for giving userspace control over ->array_sectors we need
to be able to retrieve the 'default' size, and the 'anticipated' size
when a reshape is requested.  For personalities that do not reshape emit
a warning if anything but the default size is requested.

In the raid5 case we need to update ->previous_raid_disks to make the
new 'default' size available.

Reviewed-by: Andre Noll <maan@systemlinux.org>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
15 years agomd: fix typo in FSF address
Atsushi SAKAI [Tue, 31 Mar 2009 03:57:37 +0000 (14:57 +1100)]
md: fix typo in FSF address

Hello,

 I found a typo Bosto"m" in FSF address.
And I am checking around linux source code.
Here is the only place which uses Bosto"m" (not Boston).

Signed-off-by: Atsushi SAKAI <sakaia@jp.fujitsu.com>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: add takeover support for converting raid6 back into raid5
NeilBrown [Tue, 31 Mar 2009 03:57:20 +0000 (14:57 +1100)]
md: add takeover support for converting raid6 back into raid5

If a raid6 is still in the layout that comes from converting raid5
into a raid6. this will allow us to convert it back again.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: add takeover support for raid4 -> raid5 conversion.
NeilBrown [Tue, 31 Mar 2009 03:57:09 +0000 (14:57 +1100)]
md: add takeover support for raid4 -> raid5 conversion.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: allow layout/chunksize to be changed on an active 2-drive raid5.
NeilBrown [Tue, 31 Mar 2009 03:56:41 +0000 (14:56 +1100)]
md/raid5: allow layout/chunksize to be changed on an active 2-drive raid5.

2-drive raid5's aren't very interesting.  But if you are converting
a raid1 into a raid5, you will at least temporarily have one.  And
that it a good time to set the layout/chunksize for the new RAID5
if you aren't happy with the defaults.

layout and chunksize don't actually affect the placement of data
on a 2-drive raid5, so we just do some internal book-keeping.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: add ->takeover method for raid5 to be able to take over raid1
NeilBrown [Tue, 31 Mar 2009 03:39:39 +0000 (14:39 +1100)]
md: add ->takeover method for raid5 to be able to take over raid1

The RAID1 must have two drives and be a suitable size to
be a multiple of a chunksize that isn't too small.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: add ->takeover method to support changing the personality managing an array
NeilBrown [Tue, 31 Mar 2009 03:39:39 +0000 (14:39 +1100)]
md: add ->takeover method to support changing the personality managing an array

Implement this for RAID6 to be able to 'takeover' a RAID5 array.  The
new RAID6 will use a layout which places Q on the last device, and
that device will be missing.
If there are any available spares, one will immediately have Q
recovered onto it.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: enable suspend/resume of md devices.
NeilBrown [Tue, 31 Mar 2009 03:39:39 +0000 (14:39 +1100)]
md: enable suspend/resume of md devices.

To be able to change the 'level' of an md/raid array, we need to
suspend the device so that no requests are active - then move some
pointers around etc.

The code already keeps counts of active requests and the ->quiesce
function can be used to wait until those counts hit zero.
However the quiesce function blocks new requests once they are all
ready 'inside' the personality module, and that is too late if we want
to replace the personality modules.

So make all md requests come in through a common md_make_request
function that keeps track of how many requests have entered the
modules but may not yet be on the internal reference counts.
Allow md_make_request to be blocked when we want to suspend the
device, and make it possible to wait for all those in-transit requests
to be added to internal lists so that ->quiesce can wait for them.

There is still a problem that when a request completes, we drop the
ref count inside the personality code so there is a short time between
when the refcount hits zero, and when the personality code is no
longer being used.
The personality code never blocks (schedule or spinlock) between
dropping the refcount and exiting the routine, so this should be safe
(as put_module calls synchronize_sched() before unmapping the module
code).

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: md_unregister_thread should cope with being passed NULL
NeilBrown [Tue, 31 Mar 2009 03:39:39 +0000 (14:39 +1100)]
md: md_unregister_thread should cope with being passed NULL

Mostly md_unregister_thread is only called when we know that the
thread is NULL, but sometimes we need to check first.  It is safer
to put the check inside md_unregister_thread itself.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: refactor raid5 "run"
NeilBrown [Tue, 31 Mar 2009 03:39:39 +0000 (14:39 +1100)]
md/raid5: refactor raid5 "run"

.. so that the code to create the private data structures is separate.
This will help with future code to change the level of an active
array.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: make sure new_level, new_chunksize, new_layout always have sensible values.
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md: make sure new_level, new_chunksize, new_layout always have sensible values.

When an md array is undergoing a change, we have new_* fields that
show the new values.
When no change is happening, it is least confusing if these have
the same value as the normal fields.
This is true in most cases, but not when the values are set via sysfs.

So fix this up.

A subsequent patch will BUG_ON if these things aren't consistent.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: finish support for DDF/raid6
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md/raid5: finish support for DDF/raid6

DDF requires RAID6 calculations over different devices in a different
order.
For md/raid6, we calculate over just the data devices, starting
immediately after the 'Q' block.
For ddf/raid6 we calculate over all devices, using zeros in place of
the P and Q blocks.

This requires unfortunately complex loops...

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: Add support for new layouts for raid5 and raid6.
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md/raid5: Add support for new layouts for raid5 and raid6.

DDF uses different layouts for P and Q blocks than current md/raid6
so add those that are missing.
Also add support for RAID6 layouts that are identical to various
raid5 layouts with the simple addition of one device to hold all of
the 'Q' blocks.
Finally add 'raid5' layouts to match raid4.
These last to will allow online level conversion.

Note that this does not provide correct support for DDF/raid6 yet
as the order in which data blocks are summed to produce the Q block
is significant and different between current md code and DDF
requirements.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: simplify raid5_compute_sector interface
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md/raid5: simplify raid5_compute_sector interface

Rather than passing 'pd_idx' and 'qd_idx' to be filled in, pass
a 'struct stripe_head *' and fill in the relevant fields.  This is
more extensible.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid6: remove expectation that Q device is immediately after P device.
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md/raid6: remove expectation that Q device is immediately after P device.

Code currently assumes that the devices in a raid6 stripe are
  0 1 ... N-1 P Q
in some rotated order.  We will shortly add new layouts in which
this strict pattern is broken.
So remove this expectation.  We still assume that the data disks
are roughly in-order.  However P and Q can be inserted anywhere within
that order.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: change raid5_compute_sector and stripe_to_pdidx to take a 'previous' argument
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md/raid5: change raid5_compute_sector and stripe_to_pdidx to take a 'previous' argument

This similar to the recent change to get_active_stripe.
There is no functional change, just come rearrangement to make
future patches cleaner.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd/raid5: simplify interface for init_stripe and get_active_stripe
NeilBrown [Tue, 31 Mar 2009 03:39:38 +0000 (14:39 +1100)]
md/raid5: simplify interface for init_stripe and get_active_stripe

Rather than passing 'pd_idx' and 'disks' to these functions, just pass
'previous' which tells whether to use the 'previous' or 'current'
geometry during a reshape, and let init_stripe calculate
disks and pd_idx and anything else it might need.

This is not a substantial simplification and even adds a division.
However we will shortly be adding more complexity to init_stripe
to handle more interesting 'reshape' activities, and without this
change, the interface to these functions would get very complex.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: Represent raid device size in sectors.
Andre Noll [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: Represent raid device size in sectors.

This patch renames the "size" field of struct mdk_rdev_s to
"sectors" and changes this field to store sectors instead of
blocks.

All users of this field, linear.c, raid0.c and md.c, are fixed up
accordingly which gets rid of many multiplications and divisions.

Signed-off-by: Andre Noll <maan@systemlinux.org>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: Make mddev->size sector-based.
Andre Noll [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: Make mddev->size sector-based.

This patch renames the "size" field of struct mddev_s to "dev_sectors"
and stores the number of 512-byte sectors instead of the number of
1K-blocks in it.

All users of that field, including raid levels 1,4-6,10, are adjusted
accordingly. This simplifies the code a bit because it allows to get
rid of a couple of divisions/multiplications by two.

In order to make checkpatch happy, some minor coding style issues
have also been addressed. In particular, size_store() now uses
strict_strtoull() instead of simple_strtoull().

Signed-off-by: Andre Noll <maan@systemlinux.org>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: be more consistent about setting WriteMostly flag when adding a drive to an array
NeilBrown [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: be more consistent about setting WriteMostly flag when adding a drive to an array

When a drive is added to an array using ADD_NEW_DISK, there are two
places we can get certain flags from:  the metadata on the disk or the
flags passed through the IOCTL.

For the WriteMostly flag (aka MD_DISK_WRITEMOSTLY) we take the value
from either of those sources depending on if it is set (i.e. we
effectively 'or' the two sources together).

This makes it awkward to clear, and is at best inconsistent.

As documented code (in mdadm) requires that setting
MD_DISK_WRITEMOSTLY in the ioctl will be effective, we resolve the
inconsistency by always using the value for this flag from the ioctl,
and ignoring the value on disk.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: occasionally checkpoint drive recovery to reduce duplicate effort after a crash
NeilBrown [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: occasionally checkpoint drive recovery to reduce duplicate effort after a crash

Version 1.x metadata has the ability to record the status of a
partially completed drive recovery.
However we only update that record on a clean shutdown.
It would be nice to update it on unclean shutdowns too, particularly
when using a bitmap that removes much to the 'sync' effort after an
unclean shutdown.

One complication with checkpointing recovery is that we only know
where we are up to in terms of IO requests started, not which ones
have completed.  And we need to know what has completed to record
how much is recovered.  So occasionally pause the recovery until all
submitted requests are completed, then update the record of where
we are up to.

When we have a bitmap, we already do that pause occasionally to keep
the bitmap up-to-date.  So enhance that code to record the recovery
offset and schedule a superblock update.
And when there is no bitmap, just pause 16 times during the resync to
do a checkpoint.
'16' is a fairly arbitrary number.  But we don't really have any good
way to judge how often is acceptable, and it seems like a reasonable
number for now.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: move md_k.h from include/linux/raid/ to drivers/md/
NeilBrown [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: move md_k.h from include/linux/raid/ to drivers/md/

It really is nicer to keep related code together..

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: move lots of #include lines out of .h files and into .c
NeilBrown [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: move lots of #include lines out of .h files and into .c

This makes the includes more explicit, and is preparation for moving
md_k.h to drivers/md/md.h

Remove include/raid/md.h as its only remaining use was to #include
other files.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: move most content from md.h to md_k.h
NeilBrown [Tue, 31 Mar 2009 03:33:13 +0000 (14:33 +1100)]
md: move most content from md.h to md_k.h

The extern function definitions are kernel-internal definitions, so
they belong in md_k.h

The MD_*_VERSION values could reasonably go in a number of places,
but md_u.h seems most reasonable.

This leaves almost nothing in md.h.  It will go soon.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: move LEVEL_* definition from md_k.h to md_u.h
NeilBrown [Tue, 31 Mar 2009 03:27:03 +0000 (14:27 +1100)]
md: move LEVEL_* definition from md_k.h to md_u.h

.. as they are part of the user-space interface.
Also move MdpMinorShift into there so we can remove duplication.

Lastly move mdp_major in.  It is less obviously part of the user-space
interface, but do_mounts_md.c uses it, and it is acting a bit like
user-space.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: move headers out of include/linux/raid/
Christoph Hellwig [Tue, 31 Mar 2009 03:27:03 +0000 (14:27 +1100)]
md: move headers out of include/linux/raid/

Move the headers with the local structures for the disciplines and
bitmap.h into drivers/md/ so that they are more easily grepable for
hacking and not far away.  md.h is left where it is for now as there
are some uses from the outside.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agocleanup drivers/md/Makefile
Christoph Hellwig [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
cleanup drivers/md/Makefile

Use the -y variables instead of the old -objs so we can easily add
conditional objects to the modules.  Also always use += to add
subobjects to avoid problems when placing additional objects in
some place in the file.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: stop defining MAJOR_NR
Christoph Hellwig [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
md: stop defining MAJOR_NR

MAJOR_NR was only required for magic in linux/blk.h in 2.4 or earlier
kernels, so no need to keep it around.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agoMD data integrity support
Martin K. Petersen [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
MD data integrity support

md: Add support for data integrity to MD

If all subdevices support the same protection format the MD device is
flagged as integrity capable.

Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: write bitmap information to devices that are undergoing recovery.
NeilBrown [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
md: write bitmap information to devices that are undergoing recovery.

When we add some spares to an array and start recovery, and we have
a bitmap which is stored 'internally' on all devices, we call
bitmap_write_all to make sure the bitmap is correct on the new
device(s).
However that doesn't work as write_sb_page only writes to
'In_sync' devices, and devices undergoing recovery are not
'In_sync' until recovery finishes.

So extend write_sb_page (actually next_active_rdev) to include devices
that are under recovery.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: never clear bit from the write-intent bitmap when the array is degraded.
NeilBrown [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
md: never clear bit from the write-intent bitmap when the array is degraded.

It is safe to clear a bit from the write-intent bitmap for a raid1
if we know the data has been written to all devices, which is
what the current test does.

But it is not always safe to update the 'events_cleared' counter in
that case.  This is because one request could complete successfully
after some other request has partially failed.

So simply disable the clearing and updating of events_cleared whenever
the array is degraded.  This might end up not clearing some bits that
could safely be cleared, but it is safest approach.

Note that the bug fixed here did not risk corrupting data by letting
the array get out-of-sync.  Rather it meant that when a device is
removed and re-added to the array, it might incorrectly require a full
recovery rather than just recovering based on the bitmap.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: Allow write-intent bitmaps to have chunksize < PAGE_SIZE
NeilBrown [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
md: Allow write-intent bitmaps to have chunksize < PAGE_SIZE

md currently insists that the chunk size used for write-intent
bitmaps (the amount of data that corresponds to one chunk)
be at least one page.

The reason for this restriction is lost in the mists of time,
but a review of the code (and a vague memory) suggests that the only
problem would be related to resync.  Resync tries very hard to
work in multiples of a page, but also needs to sync with units
of a bitmap_chunk too.

This connection comes out in the bitmap_start_sync call.

So change bitmap_start_sync to always work in multiples of a page.
If the bitmap chunk size is less that one page, we flag multiple
chunks as 'syncing' and generally make them all appear to the
resync routines like one chunk.

All other code either already works with data ranges that could
span multiple chunks, or explicitly only cares about a single chunk.

Signed-off-by: Neil Brown <neilb@suse.de>
15 years agomd: Fix is_mddev_idle test (again).
NeilBrown [Tue, 31 Mar 2009 03:27:02 +0000 (14:27 +1100)]
md: Fix is_mddev_idle test (again).

There are two problems with is_mddev_idle.

1/ sync_io is 'atomic_t' and hence 'int'.  curr_events and all the
   rest are 'long'.
   So if sync_io were to wrap on a 64bit host, the value of
   curr_events would go very negative suddenly, and take a very
   long time to return to positive.

   So do all calculations as 'int'.  That gives us plenty of precision
   for what we need.

2/ To initialise rdev->last_events we simply call is_mddev_idle, on
   the assumption that it will make sure that last_events is in a
   suitable range.  It used to do this, but now it does not.
   So now we need to be more explicit about initialisation.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Mon, 9 Mar 2009 20:23:59 +0000 (13:23 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/davej/cpufreq

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] Add p4-clockmod sysfs-ui removal to feature-removal schedule.
  Revert "[CPUFREQ] Disable sysfs ui for p4-clockmod."

15 years agocopy_process: fix CLONE_PARENT && parent_exec_id interaction
Oleg Nesterov [Mon, 2 Mar 2009 21:58:45 +0000 (22:58 +0100)]
copy_process: fix CLONE_PARENT && parent_exec_id interaction

CLONE_PARENT can fool the ->self_exec_id/parent_exec_id logic. If we
re-use the old parent, we must also re-use ->parent_exec_id to make
sure exit_notify() sees the right ->xxx_exec_id's when the CLONE_PARENT'ed
task exits.

Also, move down the "p->parent_exec_id = p->self_exec_id" thing, to place
two different cases together.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: David Howells <dhowells@redhat.com>
Cc: Serge E. Hallyn <serge@hallyn.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years ago[CPUFREQ] Add p4-clockmod sysfs-ui removal to feature-removal schedule.
Dave Jones [Mon, 9 Mar 2009 19:14:37 +0000 (15:14 -0400)]
[CPUFREQ] Add p4-clockmod sysfs-ui removal to feature-removal schedule.

Signed-off-by: Matthew Garrett <mjg@redhat.com>
Signed-off-by: Dave Jones <davej@redhat.com>
15 years agoRevert "[CPUFREQ] Disable sysfs ui for p4-clockmod."
Dave Jones [Mon, 9 Mar 2009 19:07:33 +0000 (15:07 -0400)]
Revert "[CPUFREQ] Disable sysfs ui for p4-clockmod."

This reverts commit e088e4c9cdb618675874becb91b2fd581ee707e6.

Removing the sysfs interface for p4-clockmod was flagged as a
regression in bug 12826.

Course of action:
 - Find out the remaining causes of overheating, and fix them
   if possible. ACPI should be doing the right thing automatically.
   If it isn't, we need to fix that.
 - mark p4-clockmod ui as deprecated
 - try again with the removal in six months.

It's not really feasible to printk about the deprecation, because
it needs to happen at all the sysfs entry points, which means adding
a lot of strcmp("p4-clockmod".. calls to the core, which.. bleuch.

Signed-off-by: Dave Jones <davej@redhat.com>
15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Mon, 9 Mar 2009 16:15:40 +0000 (09:15 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (29 commits)
  p54: fix race condition in memory management
  cfg80211: test before subtraction on unsigned
  iwlwifi: fix error flow in iwl*_pci_probe
  rt2x00 : more devices to rt73usb.c
  rt2x00 : more devices to rt2500usb.c
  bonding: Fix device passed into ->ndo_neigh_setup().
  vlan: Fix vlan-in-vlan crashes.
  net: Fix missing dev->neigh_setup in register_netdevice().
  tmspci: fix request_irq race
  pkt_sched: act_police: Fix a rate estimator test.
  tg3: Fix 5906 link problems
  SCTP: change sctp_ctl_sock_init() to try IPv4 if IPv6 fails
  IPv6: add "disable" module parameter support to ipv6.ko
  sungem: another error printed one too early
  aoe: error printed 1 too early
  net pcmcia: worklimit reaches -1
  net: more timeouts that reach -1
  net: fix tokenring license
  dm9601: new vendor/product IDs
  netlink: invert error code in netlink_set_err()
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus
Linus Torvalds [Mon, 9 Mar 2009 16:14:17 +0000 (09:14 -0700)]
Merge git://git./linux/kernel/git/rusty/linux-2.6-for-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
  lguest: fix for CONFIG_SPARSE_IRQ=y
  lguest: fix crash 'unhandled trap 13 at <native_read_msr_safe>'

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable
Linus Torvalds [Mon, 9 Mar 2009 16:13:16 +0000 (09:13 -0700)]
Merge git://git./linux/kernel/git/mason/btrfs-unstable

* git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  Btrfs: fix spinlock assertions on UP systems

15 years agoBtrfs: fix spinlock assertions on UP systems
Chris Mason [Mon, 9 Mar 2009 15:45:38 +0000 (11:45 -0400)]
Btrfs: fix spinlock assertions on UP systems

btrfs_tree_locked was being used to make sure a given extent_buffer was
properly locked in a few places.  But, it wasn't correct for UP compiled
kernels.

This switches it to using assert_spin_locked instead, and renames it to
btrfs_assert_tree_locked to better reflect how it was really being used.

Signed-off-by: Chris Mason <chris.mason@oracle.com>
15 years agoFix fixpoint divide exception in acct_update_integrals
Heiko Carstens [Mon, 9 Mar 2009 12:31:59 +0000 (13:31 +0100)]
Fix fixpoint divide exception in acct_update_integrals

Frans Pop reported the crash below when running an s390 kernel under Hercules:

  Kernel BUG at 000738b4  verbose debug info unavailable!
  fixpoint divide exception: 0009  #1! SMP
  Modules linked in: nfs lockd nfs_acl sunrpc ctcm fsm tape_34xx
     cu3088 tape ccwgroup tape_class ext3 jbd mbcache dm_mirror dm_log dm_snapshot
     dm_mod dasd_eckd_mod dasd_mod
  CPU: 0 Not tainted 2.6.27.19 #13
  Process awk (pid: 2069, task: 0f9ed9b8, ksp: 0f4f7d18)
  Krnl PSW : 070c1000 800738b4 (acct_update_integrals+0x4c/0x118)
             R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:0 CC:1 PM:0
  Krnl GPRS: 00000000 000007d0 7fffffff fffff830
             00000000 ffffffff 00000002 0f9ed9b8
             00000000 00008ca0 00000000 0f9ed9b8
             0f9edda4 8007386e 0f4f7ec8 0f4f7e98
  Krnl Code: 800738aaa71807d0         lhi     %r1,2000
             800738ae8c200001         srdl    %r2,1
             800738b2: 1d21             dr      %r2,%r1
            >800738b45810d10e         l       %r1,270(%r13)
             800738b8: 1823             lr      %r2,%r3
             800738ba4130f060         la      %r3,96(%r15)
             800738be: 0de1             basr    %r14,%r1
             800738c05800f060         l       %r0,96(%r15)
  Call Trace:
  ( <000000000004fdea>! blocking_notifier_call_chain+0x1e/0x2c)
    <0000000000038502>! do_exit+0x106/0x7c0
    <0000000000038c36>! do_group_exit+0x7a/0xb4
    <0000000000038c8e>! SyS_exit_group+0x1e/0x30
    <0000000000021c28>! sysc_do_restart+0x12/0x16
    <0000000077e7e924>! 0x77e7e924

Reason for this is that cpu time accounting usually only happens from
interrupt context, but acct_update_integrals gets also called from
process context with interrupts enabled.

So in acct_update_integrals we may end up with the following scenario:

Between reading tsk->stime/tsk->utime and tsk->acct_timexpd an interrupt
happens which updates accouting values.  This causes acct_timexpd to be
greater than the former stime + utime.  The subsequent calculation of

dtime = cputime_sub(time, tsk->acct_timexpd);

will be negative and the division performed by

cputime_to_jiffies(dtime)

will generate an exception since the result won't fit into a 32 bit
register.

In order to fix this just always disable interrupts while accessing any
of the accounting values.

Reported by: Frans Pop <elendil@planet.nl>
Tested by: Frans Pop <elendil@planet.nl>
Cc: stable@kernel.org
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agolguest: fix for CONFIG_SPARSE_IRQ=y
Rusty Russell [Mon, 9 Mar 2009 16:06:28 +0000 (10:06 -0600)]
lguest: fix for CONFIG_SPARSE_IRQ=y

Impact: remove lots of lguest boot WARN_ON() when CONFIG_SPARSE_IRQ=y

We now need to call irq_to_desc_alloc_cpu() before
set_irq_chip_and_handler_name(), but we can't do that from init_IRQ (no
kmalloc available).

So do it as we use interrupts instead.  Also means we only alloc for
irqs we use, which was the intent of CONFIG_SPARSE_IRQ anyway.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Ingo Molnar <mingo@redhat.com>
15 years agolguest: fix crash 'unhandled trap 13 at <native_read_msr_safe>'
Rusty Russell [Mon, 9 Mar 2009 16:06:22 +0000 (10:06 -0600)]
lguest: fix crash 'unhandled trap 13 at <native_read_msr_safe>'

Impact: fix lguest boot crash on modern Intel machines

The code in early_init_intel does:

if (c->x86 > 6 || (c->x86 == 6 && c->x86_model >= 0xd)) {
u64 misc_enable;

rdmsrl(MSR_IA32_MISC_ENABLE, misc_enable);

And that rdmsr faults (not allowed from non-0 PL).  We can get around
this by mugging the family ID part of the cpuid.  5 seems like a good
number.

Of course, this is a hack (how very lguest!).  We could just indicate
that we don't support MSRs, or implement lguest_rdmst.

Reported-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Tested-by: Patrick McHardy <kaber@trash.net>
15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc
Linus Torvalds [Sun, 8 Mar 2009 17:37:57 +0000 (10:37 -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:
  mmc: fix data timeout for SEND_EXT_CSD

15 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 8 Mar 2009 17:30:18 +0000 (10:30 -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:
  rcu: increment quiescent state counter in ksoftirqd()

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 8 Mar 2009 17:27:13 +0000 (10:27 -0700)]
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, pebs: correct qualifier passed to ds_write_config() from ds_request_pebs()
  x86, bts: remove bad warning
  x86: add Dell XPS710 reboot quirk
  x86, math-emu: fix init_fpu for task != current
  x86: EFI: Back efi_ioremap with init_memory_mapping instead of FIX_MAP
  x86: fix DMI on EFI

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Sun, 8 Mar 2009 17:25:13 +0000 (10:25 -0700)]
Merge git://git./linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  [WATCHDOG] orion5x_wdt.c: 'ORION5X_TCLK' undeclared
  [WATCHDOG] gef_wdt.c: fsl_get_sys_freq() failure not noticed
  [WATCHDOG] ks8695_wdt.c: 'CLOCK_TICK_RATE' undeclared
  [WATCHDOG] rc32434_wdt: fix sections
  [WATCHDOG] rc32434_wdt: fix watchdog driver

15 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Linus Torvalds [Sun, 8 Mar 2009 17:24:57 +0000 (10:24 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: fix ext4_free_inode() vs. ext4_claim_inode() race

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney...
Linus Torvalds [Sun, 8 Mar 2009 17:24:39 +0000 (10:24 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/cooloney/blackfin-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/blackfin-2.6: (28 commits)
  Blackfin arch: SPI_MMC is now mainlined MMC_SPI
  Blackfin arch: disable legacy /proc/scsi/ support by default
  Blackfin arch: remove duplicated ANOMALY_05000448 ifdef check
  Blackfin arch: add stubs for anomalies 447 and 448
  Blackfin arch: cleanup bfin_sport.h header and export it to userspace
  Blackfin arch: fix bug - gdb signull case make trunk kernel panic frequently
  Blackfin arch: remove spurious dash when dcache is off
  Blackfin arch: mark init_pda as __init as only __init funcs all it
  Blackfin arch: fix bug - On bf548-ezkit, ethernet fails to work after wakeup from "mem"
  Blackfin arch: Random read/write errors are a bad thing
  Blackfin arch: update default kernel config, select KSZ8893M driver for BF518
  Blackfin arch: Fix bug - KGDB single step into the middle of a 4 bytes instruction on bf561 after soft bp is hit
  Blackfin arch: Fix bug - make ksz8893m driver available when bfin_mac is enabled
  Blackfin arch: make sure people do not set the kernel load address too high
  Blackfin arch: fix bug - The SPORT_HYS bit is not set for BF561 0.5
  Blackfin arch: update anomaly sheets to match latest public info
  Blackfin arch: Fix BUG - kernel fails to build in pm.c when allow wakeup fromi standby by GPIO
  Blackfin arch: PM_BFIN_WAKE_GP: update help
  Blackfin arch: fix bug - kgdb fails to continue after setting breakpoint on bf561-ezkit kernel with smp patch
  Blackfin arch: Enable Write Back Cache on all Blackfin Boards
  ...

15 years agoMerge branch 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx
Linus Torvalds [Sun, 8 Mar 2009 17:23:05 +0000 (10:23 -0700)]
Merge branch 'fixes' of git://git./linux/kernel/git/djbw/async_tx

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx:
  dmatest: fix use after free in dmatest_exit
  ipu_idmac: fix spinlock type
  iop-adma, mv_xor: fix mem leak on self-test setup failure
  fsldma: fix off by one in dma_halt
  I/OAT: fail self-test if callback test reaches timeout
  I/OAT: update driver version and copyright dates
  I/OAT: list usage cleanup
  I/OAT: set tcp_dma_copybreak to 256k for I/OAT ver.3
  I/OAT: cancel watchdog before dma remove
  I/OAT: fail initialization on zero channels detection
  I/OAT: do not set DCACTRL_CMPL_WRITE_ENABLE for I/OAT ver.3
  I/OAT: add verification for proper APICID_TAG_MAP setting by BIOS
  dmaengine: update kerneldoc

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6
Linus Torvalds [Sun, 8 Mar 2009 17:22:22 +0000 (10:22 -0700)]
Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6:
  ata: add CFA specific identify data words
  remove stale comment from <linux/hdreg.h>
  AT91: initialize Compact Flash on AT91SAM9263 cpu
  ide: add at91_ide driver
  ide: allow to wrap interrupt handler
  ide-iops: fix odd-length ATAPI PIO transfers
  ide: NULL noise: drivers/ide/ide-*.c
  ide: expiry() returns int, negative expiry() return values won't be noticed

15 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Sun, 8 Mar 2009 17:22:01 +0000 (10:22 -0700)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  libata: Don't trust current capacity values in identify words 57-58
  libata: make sure port is thawed when skipping resets
  sata_nv: fix module parameter description
  ahci: Add the Device IDs for MCP89 and remove IDs of MCP7B to/from ahci.c
  libata: don't use on-stack sense buffer
  libata: align ap->sector_buf
  libata: fix dma_unmap_sg misuse
  libata: change drive ready wait after hard reset to 5s

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/pkl/squashfs-linus
Linus Torvalds [Sun, 8 Mar 2009 17:21:31 +0000 (10:21 -0700)]
Merge git://git./linux/kernel/git/pkl/squashfs-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/pkl/squashfs-linus:
  Squashfs: frag_size should be signed, as it can hold an error result
  Squashfs: fix documentation typo, Cramfs filesystem limit is 256 MiB
  Squashfs: Fix oops when reading fsfuzzer corrupted filesystems

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris...
Linus Torvalds [Sun, 8 Mar 2009 17:21:10 +0000 (10:21 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jmorris/security-testing-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jmorris/security-testing-2.6:
  smack: fixes for unlabeled host support

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input
Linus Torvalds [Sun, 8 Mar 2009 17:14:19 +0000 (10:14 -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: serio - fix protocol number for TouchIT213

15 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Sun, 8 Mar 2009 17:13:28 +0000 (10:13 -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 PCI DMA flag propagation on SN (Altix) with PICs

15 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Sun, 8 Mar 2009 17:08:57 +0000 (10:08 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  block: fix missing bio back/front segment size setting in blk_recount_segments()
  loop: don't increment p->offset with (size_t) -EINVAL
  cciss: remove 30 second initial timeout on controller reset
  Fix kernel NULL pointer dereference in xen-blkfront

15 years agoMerge branch 'fix/hda' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Sun, 8 Mar 2009 17:03:31 +0000 (10:03 -0700)]
Merge branch 'fix/hda' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'fix/hda' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda - Fix headphone-detect regression with multiple HP jacks
  ALSA: hda - Fix typos in slave controls in patch_sigmatel.c

15 years agoMIPS: compat: Implement is_compat_task.
Ralf Baechle [Thu, 5 Mar 2009 10:45:48 +0000 (11:45 +0100)]
MIPS: compat: Implement is_compat_task.

This is a build fix required after "x86-64: seccomp: fix 32/64 syscall
hole" (commit 5b1017404aea6d2e552e991b3fd814d839e9cd67).  MIPS doesn't
have the issue that was fixed for x86-64 by that patch.

This also doesn't solve the N32 issue which is that N32 seccomp processes
will be treated as non-compat processes thus only have access to N64
syscalls.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agommc: fix data timeout for SEND_EXT_CSD
Adrian Hunter [Tue, 10 Feb 2009 14:32:33 +0000 (16:32 +0200)]
mmc: fix data timeout for SEND_EXT_CSD

Commit 0d3e0460f307e84904968aad6cff97bd688583d8
"MMC: CSD and CID timeout values" inadvertently broke
the timeout for the MMC command SEND_EXT_CSD.

This patch puts it back again.

Depending on the characteristics of the controller,
this bug may prevent the use of MMC cards.

Signed-off-by: Adrian Hunter <adrian.hunter@nokia.com>
Signed-off-by: Pierre Ossman <drzeus@drzeus.cx>
15 years agoInput: serio - fix protocol number for TouchIT213
Dmitry Torokhov [Sat, 7 Mar 2009 21:39:22 +0000 (13:39 -0800)]
Input: serio - fix protocol number for TouchIT213

Protocol 0x37 has been reserved for iNexio devices and Sahara
was supposed to get 0x38.

Reported-by: Claudio Nieder <private@claudio.ch>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
15 years agop54: fix race condition in memory management
Christian Lamparter [Thu, 5 Mar 2009 23:53:59 +0000 (00:53 +0100)]
p54: fix race condition in memory management

This patch fixes a number of race conditions in the driver.
Up until now, "entry" pointer was initialized before acquiring the right lock.

Signed-off-by: Christian Lamparter <chunkeey@web.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agocfg80211: test before subtraction on unsigned
Roel Kluin [Tue, 3 Mar 2009 21:55:21 +0000 (22:55 +0100)]
cfg80211: test before subtraction on unsigned

freq_diff is unsigned, so test before subtraction

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years ago[IA64] fix PCI DMA flag propagation on SN (Altix) with PICs
Jeremy Higdon [Wed, 4 Mar 2009 20:09:46 +0000 (12:09 -0800)]
[IA64] fix PCI DMA flag propagation on SN (Altix) with PICs

We recently discovered a problem with passing of DMA attributes on SN
systems with the older PIC chips.

[akpm@linux-foundation.org: coding-style fixes]

Signed-off-by: Jeremy Higdon <jeremy@sgi.com>
Cc: <habeck@sgi.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Tony Luck <tony.luck@intel.com>
15 years agox86, pebs: correct qualifier passed to ds_write_config() from ds_request_pebs()
Markus Metzger [Thu, 5 Mar 2009 07:57:21 +0000 (08:57 +0100)]
x86, pebs: correct qualifier passed to ds_write_config() from ds_request_pebs()

ds_write_config() can write the BTS as well as the PEBS part of
the DS config. ds_request_pebs() passes the wrong qualifier, which
results in the wrong configuration to be written.

Reported-by: Stephane Eranian <eranian@googlemail.com>
Signed-off-by: Markus Metzger <markus.t.metzger@intel.com>
LKML-Reference: <20090305085721.A22550@sedona.ch.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, bts: remove bad warning
Markus Metzger [Thu, 5 Mar 2009 07:49:54 +0000 (08:49 +0100)]
x86, bts: remove bad warning

In case a ptraced task is reaped (while the tracer is still attached),
ds_exit_thread() is called before ptrace_exit(). The latter will
release the bts_tracer and remove the thread's ds_ctx.
The former will WARN() if the context is not NULL.

Oleg Nesterov submitted patches that move ptrace_exit() before
exit_thread() and thus reverse the order of the above calls.

Remove the bad warning. I will add it again when Oleg's changes are in.

Signed-off-by: Markus Metzger <markus.t.metzger@intel.com>
LKML-Reference: <20090305084954.A22000@sedona.ch.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoALSA: hda - Fix headphone-detect regression with multiple HP jacks
Takashi Iwai [Fri, 6 Mar 2009 08:43:58 +0000 (09:43 +0100)]
ALSA: hda - Fix headphone-detect regression with multiple HP jacks

The recent changes over the DAC detection mechanism in patch_sigmatel.c
breaks the HP detection on the machines with multiple HP jacks.
It's basically because of the workaround to support the multi-channel
output.  Since the HP detection is more important feature, disable
the HP-swap workaroud temporarily.

Reference: Novell bnc#482052
https://bugzilla.novell.com/show_bug.cgi?id=482052

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: hda - Fix typos in slave controls in patch_sigmatel.c
Takashi Iwai [Fri, 6 Mar 2009 08:42:07 +0000 (09:42 +0100)]
ALSA: hda - Fix typos in slave controls in patch_sigmatel.c

"Headphone Playback ..." appears twice in slave_vols[] and slave_sws[].
They should be "Headphone Playback2 ..."

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoblock: fix missing bio back/front segment size setting in blk_recount_segments()
Jens Axboe [Fri, 6 Mar 2009 07:55:24 +0000 (08:55 +0100)]
block: fix missing bio back/front segment size setting in blk_recount_segments()

Commit 1e42807918d17e8c93bf14fbb74be84b141334c1 introduced a bug where we
don't get front/back segment sizes in the bio in blk_recount_segments().
Fix this by tracking the back bio as well as the front bio in
__blk_recalc_rq_segments(), this also cleans up the interface by getting
rid of the segment size pointer passing.

Tested-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years ago[WATCHDOG] orion5x_wdt.c: 'ORION5X_TCLK' undeclared
Wim Van Sebroeck [Wed, 25 Feb 2009 12:33:01 +0000 (13:33 +0100)]
[WATCHDOG] orion5x_wdt.c: 'ORION5X_TCLK' undeclared

orion5x_wdt no longer compiled after the changes in commit
ebe35aff883496c07248df82c8576c3b6e84bbbe ("Orion: prepare for
runtime-determined timer tick rate").

The tick rate define (ORION5X_TCLK) was removed in favor of a runtime
detection. The quick fix is to add the define in the watchdog driver.
The fix is not correct for all supported orion5x platforms, but since
the supported platforms right now are 133 Mhz and 166 Mhz, it won't
be _that_ far off. ;-) A fix that uses the runtime-determined timer
tick rate will be applied later.

Signed-off-by: Wim Van Sebroeck <wim@iguana.be>
Cc: Kristof Provost <kristof@sigsegv.be>
Acked-by: Lennert Buytenhek <buytenh@wantstofly.org>
Cc: Nicolas Pitre <nico@cam.org>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: Andrew Morton <akpm@linux-foundation.org>
15 years agoiwlwifi: fix error flow in iwl*_pci_probe
Reinette Chatre [Tue, 3 Mar 2009 19:37:04 +0000 (11:37 -0800)]
iwlwifi: fix error flow in iwl*_pci_probe

Both the agn and 3945 drivers has some problems with dealing with
errors in their probe functions. Ensure that a goto will undo only
things that was done before the goto was called.

Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agort2x00 : more devices to rt73usb.c
Xose Vazquez Perez [Fri, 27 Feb 2009 23:34:23 +0000 (00:34 +0100)]
rt2x00 : more devices to rt73usb.c

add more usb_dev to rt73usb.c . IDs 'stolen' from the
windows inf file(10/21/2008, 1.03.02.0000) plus some
from the Ralink linux driver(2009_0206_RT73_Linux_STA_Drv1.1.0.2.tar.bz2)

Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agort2x00 : more devices to rt2500usb.c
Xose Vazquez Perez [Fri, 27 Feb 2009 23:26:09 +0000 (00:26 +0100)]
rt2x00 : more devices to rt2500usb.c

add more usb_dev to rt2500usb.c . IDs 'stolen' from the
windows inf file(02/12/2009, 2.01.01.0015).

Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
15 years agoBlackfin arch: SPI_MMC is now mainlined MMC_SPI
Michael Hennerich [Thu, 5 Mar 2009 16:27:57 +0000 (00:27 +0800)]
Blackfin arch: SPI_MMC is now mainlined MMC_SPI

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
15 years agoBlackfin arch: disable legacy /proc/scsi/ support by default
Mike Frysinger [Thu, 5 Mar 2009 16:27:17 +0000 (00:27 +0800)]
Blackfin arch: disable legacy /proc/scsi/ support by default

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
15 years agoBlackfin arch: remove duplicated ANOMALY_05000448 ifdef check
Mike Frysinger [Thu, 5 Mar 2009 16:24:01 +0000 (00:24 +0800)]
Blackfin arch: remove duplicated ANOMALY_05000448 ifdef check

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
15 years agoata: add CFA specific identify data words
Sergei Shtylyov [Thu, 5 Mar 2009 16:20:55 +0000 (17:20 +0100)]
ata: add CFA specific identify data words

Declare CFA specific identify data words 162 and 163 for future use.

Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>
[bart: update patch summary/description]
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoBlackfin arch: add stubs for anomalies 447 and 448
Mike Frysinger [Thu, 5 Mar 2009 16:20:49 +0000 (00:20 +0800)]
Blackfin arch: add stubs for anomalies 447 and 448

Signed-off-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Bryan Wu <cooloney@kernel.org>
15 years agoremove stale comment from <linux/hdreg.h>
Bartlomiej Zolnierkiewicz [Thu, 5 Mar 2009 15:10:59 +0000 (16:10 +0100)]
remove stale comment from <linux/hdreg.h>

HDIO_GET_IDENTITY returns 256 words currently.

Noticed by Norman Diamond.

Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoAT91: initialize Compact Flash on AT91SAM9263 cpu
Stanislaw Gruszka [Thu, 5 Mar 2009 15:10:58 +0000 (16:10 +0100)]
AT91: initialize Compact Flash on AT91SAM9263 cpu

Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Cc: Andrew Victor <avictor.za@gmail.com>
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Acked-by: Andrew Victor <linux@maxim.org.za>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoide: add at91_ide driver
Stanislaw Gruszka [Thu, 5 Mar 2009 15:10:58 +0000 (16:10 +0100)]
ide: add at91_ide driver

This is IDE host driver for AT91 (SAM9, CAP9, AT572D940HF) Static Memory
Controller with Compact Flash True IDE Mode logic.

Driver have to switch 8/16 bit bus width when accessing Task Tile or Data
Register. Moreover some extra things need to be done when setting PIO mode.
Only PIO mode is used, hardware have no DMA support. If interrupt line is
connected through GPIO extra quirk is needed to cope with fake interrupts.

Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Cc: Andrew Victor <avictor.za@gmail.com>
Acked-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoide: allow to wrap interrupt handler
Stanislaw Gruszka [Thu, 5 Mar 2009 15:10:57 +0000 (16:10 +0100)]
ide: allow to wrap interrupt handler

Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Cc: Andrew Victor <linux@maxim.org.za>
[bart: minor checkpatch.pl / CodingStyle fixups]
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoide-iops: fix odd-length ATAPI PIO transfers
Sergei Shtylyov [Thu, 5 Mar 2009 15:10:56 +0000 (16:10 +0100)]
ide-iops: fix odd-length ATAPI PIO transfers

Commit 9567b349f7e7dd7e2483db99ee8e4a6fe0caca38 (ide: merge ->atapi_*put_bytes
and ->ata_*put_data methods) introduced a regression  WRT the odd-length ATAPI
PIO transfers -- the final word didn't get written (causing command timeouts).

Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
15 years agoide: NULL noise: drivers/ide/ide-*.c
Hannes Eder [Thu, 5 Mar 2009 15:10:56 +0000 (16:10 +0100)]
ide: NULL noise: drivers/ide/ide-*.c

Fix this sparse warnings:
  drivers/ide/ide-disk_proc.c:130:11: warning: Using plain integer as NULL pointer
  drivers/ide/ide-floppy_proc.c:32:11: warning: Using plain integer as NULL pointer
  drivers/ide/ide-proc.c:234:11: warning: Using plain integer as NULL pointer
  drivers/ide/ide-tape.c:2141:11: warning: Using plain integer as NULL pointer

Signed-off-by: Hannes Eder <hannes@hanneseder.net>
Cc: trivial@kernel.org
Cc: kernel-janitors@vger.kernel.org
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>