This reverts commit 1f84260c, which is
suspected to be the reason for some very occasional and hard-to-trigger
crashes that usually look related to memory allocation (mostly reported
in networking, but since that's generally the most common source of
shortlived allocations - and allocations in interrupt contexts - that in
itself is not a big clue).

See for example
	http://bugzilla.kernel.org/show_bug.cgi?id=9973
	http://lkml.org/lkml/2008/2/19/278
etc.

One promising suspicion for what the root cause of bug is (which also
explains why it's so hard to trigger in practice) came from Eric
Dumazet:

   "I wonder how SLUB_FASTPATH is supposed to work, since it is affected
    by a classical ABA problem of lockless algo.

    cmpxchg_local(&c->freelist, object, object[c->offset]) can succeed,
    while an interrupt came (on this cpu), and several allocations were
    done, and one free was performed at the end of this interruption, so
    'object' was recycled.

    c->freelist can then contain the previous value (object), but
    object[c->offset] was changed by IRQ.

    We then put back in freelist an already allocated object."

but another reason for the revert is simply that everybody agrees that
this code was the main suspect just by virtue of the pattern of oopses.

Cc: Torsten Kaiser <just.for.lkml@googlemail.com>
Cc: Christoph Lameter <clameter@sgi.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

* 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6:
  [S390] qdio: FCP/SCSI write I/O stagnates on LPAR
  [S390] Fix futex_atomic_cmpxchg_std inline assembly.
  [S390] dcss: Fix Unlikely(x) != y
  [S390] sclp: clean up send/receive naming scheme
  [S390] etr: fix compile error on !SMP
  [S390] qdio: fix qdio_activate timeout handling.
  [S390] Initialize per cpu lowcores on cpu hotplug.
  [S390] find bit corner case.
  [S390] dasd: fix locking in __dasd_device_process_final_queue
  [S390] Make sure enabled wait psw is loaded in default_idle.
  [S390] Let NR_CPUS default to 32/64 on s390/s390x.
  [S390] cio: Do timed recovery on workqueue.
  [S390] cio: Remember to initialize recovery_lock.

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  libata: implement drain buffers
  libata: eliminate the home grown dma padding in favour of
  block: clear drain buffer if draining for write command
  block: implement request_queue->dma_drain_needed
  block: add request->raw_data_len
  block: update bio according to DMA alignment padding
  libata: update ATAPI overflow draining
  elevator: make elevator_get() attempt to load the appropriate module
  cfq-iosched: add hlist for browsing parallel to the radix tree
  block: make blk_rq_map_user() clear ->bio if it unmaps it
  fs/block_dev.c: remove #if 0'ed code
  make struct def_blk_aops static
  make blk_settings_init() static
  make blk_ioc_init() static
  make blk-core.c:request_cachep static again

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  [WATCHDOG] HP ProLiant WatchDog driver
  [WATCHDOG] blackfin Watchdog driver: relocate all strings used in __init functions to __initdata
  [WATCHDOG] Convert mtx1 wdt to be a platform device and use generic GPIO API
  [WATCHDOG] Add support for SB1 hardware watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/tglx/linux-2.6-genirq:
  genirq: do not leave interupts enabled on free_irq
  genirq: spurious.c: use time_* macros

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Add regs_return_value().
  [SPARC64]: Kill pcic_present().
  [SPARC]: Kill 'prom_palette'.
  [ATYFB]: Kill 'prom_palette' sparc code.
  [SPARC64]: Kill 'prom_keyboard'.
  [SPARC]: Kill extern decl of 'panic_setup'.
  [SPARC64]: Delete 'boot_flags'.
  [SPARC64]: Kill unused function 'kernel_enter_debugger'.
  [SPARC64] arch/sparc64/kernel/unaligned.c: Use time_* macros
  [SPARC64]: Always register a PROM based early console.
  [SPARC64]: Update defconfig.
  [SPARC64]: Add -mtune=ultrasparc3 if possible.
  [SPARC64]: Remove Makefile code for ancient gcc and binutils.
  [SPARC64]: Remove DEBUG_BOOTMEM.
  [SPARC64]: Use shorter "get_zeroed_page" call.
  [SPARC]: Use shorter form of "get_zeroed_page".
  [SPARC]: video/cg14.c and video/sbuslib.c build fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (60 commits)
  [NIU]: Bump driver version and release date.
  [NIU]: Fix BMAC alternate MAC address indexing.
  net: fix kernel-doc warnings in header files
  [IPV6]: Use BUG_ON instead of if + BUG in fib6_del_route.
  [IPV6]: dst_entry leak in ip4ip6_err. (resend)
  bluetooth: do not move child device other than rfcomm
  bluetooth: put hci dev after del conn
  [NET]: Elminate spurious print_mac() calls.
  [BLUETOOTH] hci_sysfs.c: Kill build warning.
  [NET]: Remove MAC_FMT
  net/8021q/vlan_dev.c: Use print_mac.
  [XFRM]: Fix ordering issue in xfrm_dst_hash_transfer().
  [BLUETOOTH] net/bluetooth/hci_core.c: Use time_* macros
  [IPV6]: Fix hardcoded removing of old module code
  [NETLABEL]: Move some initialization code into __init section.
  [NETLABEL]: Shrink the genl-ops registration code.
  [AX25] ax25_out: check skb for NULL in ax25_kick()
  [TCP]: Fix tcp_v4_send_synack() comment
  [IPV4]: fix alignment of IP-Config output
  Documentation: fix tcp.txt
  ...

Smack uses CIPSO labeling, but allows for unlabeled packets by
specifying an "ambient" label that is applied to incoming unlabeled
packets.

Because the other end of the connection may dislike IP options, and ssh
is one know application that behaves thus, it is prudent to respond in
kind.

This patch changes the network labeling behavior such that an outgoing
packet that would be given a CIPSO label that matches the ambient label
is left unlabeled.  An "unlbl" domain is added and the netlabel
defaulting mechanism invoked rather than assuming that everything is
CIPSO.  Locking has been added around changes to the ambient label as
the mechanisms used to do so are more involved.
Signed-off-by: NCasey Schaufler <casey@schaufler-ca.com>
Acked-by: NPaul Moore <paul.moore@hp.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

If running on LPAR, qdio might overlook an incoming buffer in certain
scenarios. The patch makes sure that incoming buffers are detected
immediately in all situations.
Signed-off-by: NUrsula Braun <braunu@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Add missing exception table entry so that the kernel can handle
proctection exceptions as well on the cs instruction. Currently only
specification exceptions are handled correctly.
The missing entry allows user space to crash the kernel.

Cc: stable <stable@kernel.org>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix Unlikely(x) != y

Cc: Gerald Schaefer <geraldsc@de.ibm.com>
Cc: Stefan Weinhuber <wein@de.ibm.com>
Cc: Carsten Otte <cotte@de.ibm.com>
Signed-off-by: NRoel Kluin <12o3l@tiscali.nl>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Make state change events adjust the correct mask by cleaning up
naming inconsistencies. Also remove chance for lockup by removing
unnecessary mask related check before reading events.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Since a5fbb6d1
"KVM: fix !SMP build error" smp_call_function isn't a define anymore
that folds into nothing but a define that calls up_smp_call_function
with all parameters. Hence we cannot #ifdef out the unused code
anymore...
This seems to be the preferred method, so do this for s390 as well.

arch/s390/kernel/time.c: In function 'etr_sync_clock':
arch/s390/kernel/time.c:825: error: 'clock_sync_cpu_start' undeclared
arch/s390/kernel/time.c:862: error: 'clock_sync_cpu_end' undeclared
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Current code in qdio_activate waits for at least 5 seconds
until it returns. It may return earlier if an error occurs,
but not if everything is ok. This large timeout value
became visible with commit dfa77f61
"qdio: set QDIO_ACTIVATE_TIMEOUT to 5s", which intended to
fix the timeout value which was zero. In turn setting an
FCP adapter online took 5 seconds.

In practice waiting for 5ms before continuing is sufficient
as pointed out by Utz Bacher and Cornelia Huck.

Cc: Utz Bacher <utz.bacher@de.ibm.com>
Cc: Jan Glauber <jan.glauber@de.ibm.com>
Cc: Ursula Braun <braunu@de.ibm.com>
Cc: Martin Peschke <mp3@de.ibm.com>
Acked-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Just copy the first 512 read-only bytes of the current cpu lowcore if
a new cpu gets onlined. The rest is zeroed out and must be explicitly
initialized. Current code just copies the entire lowcore and
initializes the needed fields.
This should reveal bugs in future enhancements quite early.
Also when the lowcore of the first cpu is replaced this is now done
atomically (no interrupts, no machine checks).
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix [ext2_]find_first_[zero_]bit for the corner case of an all clear
or all set bit field by always handling that last word of the bit field
with __ffz_word/__ffs_word.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

After setting the status of the cqr and releasing the lock for the
block cqr queue, we call the cqr callback function, which will usually
just trigger the dasd_block_tasklet. But when the tasklet is already
running the cqr might be processed before we invoke the callback
function. In rare cases the callback pointer may already be invalid
by the time we want to call it, which will result in a panic.
Solution: Call the callback function first and then release the lock.
Signed-off-by: NStefan Weinhuber <wein@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

If both NO_IDLE_HZ and VIRT_TIMER are disabled default_idle won't load
an enabled wait psw and busy loop instead. This is because the
idle_chain is empty and the return value of atomic_notifier_call_chain
will be NOTIFY_DONE, which causes default_idle to return instead of
loading an enabled wait psw.
Fix this by calling __atomic_notifier_call_chain instead and add proper
return value handling.
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

We can't do our recovery in softirq context, so we schedule it from
our timer function.
Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Signed-off-by: NCornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

This just updates the libata slave configure routine to take advantage
of the block layer drain buffers.  It also adjusts the size lengths in
the atapi code to add the drain buffer to the DMA length so the driver
knows it can rely on it.

I suspect I should also be checking for AHCI as well as ATA_DEV_ATAPI,
but I couldn't see how to do that easily.

tj: * atapi_drain_needed() added such that draining is applied to only
      misc ATAPI commands.
    * q->bounce_gfp used when allocating drain buffer.
    * Now duplicate ATAPI PIO drain logic dropped.
    * ata_dev_printk() used instead of sdev_printk().
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

that provided by the block layer

ATA requires that all DMA transfers begin and end on word boundaries.
Because of this, a large amount of machinery grew up in ide to adjust
scatterlists on this basis.  However, as of 2.5, the block layer has a
dma_alignment variable which ensures both the beginning and length of a
DMA transfer are aligned on the dma_alignment boundary.  Although the
block layer does adjust the beginning of the transfer to ensure this
happens, it doesn't actually adjust the length, it merely makes sure
that space is allocated for transfers beyond the declared length.  The
upshot of this is that scatterlists may be padded to any size between
the actual length and the length adjusted to the dma_alignment safely
knowing that memory is allocated in this region.

Right at the moment, SCSI takes the default dma_aligment which is on a
512 byte boundary.  Note that this aligment only applies to transfers
coming in from user space.  However, since all kernel allocations are
automatically aligned on a minimum of 32 byte boundaries, it is safe to
adjust them in this manner as well.

tj: * Adjusting sg after padding is done in block layer.  Make libata
      set queue alignment correctly for ATAPI devices and drop broken
      sg mangling from ata_sg_setup().
    * Use request->raw_data_len for ATAPI transfer chunk size.
    * Killed qc->raw_nbytes.
    * Separated out killing qc->n_iter.
Signed-off-by: NJames Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Clear drain buffer before chaining if the command in question is a
write.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Draining shouldn't be done for commands where overflow may indicate
data integrity issues.  Add dma_drain_needed callback to
request_queue.  Drain buffer is appened iff this function returns
non-zero.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

With padding and draining moved into it, block layer now may extend
requests as directed by queue parameters, so now a request has two
sizes - the original request size and the extended size which matches
the size of area pointed to by bios and later by sgs.  The latter size
is what lower layers are primarily interested in when allocating,
filling up DMA tables and setting up the controller.

Both padding and draining extend the data area to accomodate
controller characteristics.  As any controller which speaks SCSI can
handle underflows, feeding larger data area is safe.

So, this patch makes the primary data length field, request->data_len,
indicate the size of full data area and add a separate length field,
request->raw_data_len, for the unmodified request size.  The latter is
used to report to higher layer (userland) and where the original
request size should be fed to the controller or device.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

DMA start address and transfer size alignment for PC requests are
achieved using bio_copy_user() instead of bio_map_user().  This works
because bio_copy_user() always uses full pages and block DMA alignment
isn't allowed to go over PAGE_SIZE.

However, the implementation didn't update the last bio of the request
to make this padding visible to lower layers.  This patch makes
blk_rq_map_user() extend the last bio such that it includes the
padding area and the size of area pointed to by the request is
properly aligned.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

For misc ATAPI commands which transfer variable length data to the
host, overflow can occur due to application or hardware bug.  Such
overflows can be ignored safely as long as overflow data is properly
drained.  libata HSM implementation has this implemented in
__atapi_pio_bytes() and recently updated for 2.6.24-rc but it requires
further improvements.  Improve drain logic such that...

* Report overflow errors using ehi desc mechanism instead of printing
  directly.

* Properly calculate the number of bytes to be drained considering
  actual number of consumed bytes for partial draining.
Signed-off-by: NTejun Heo <htejun@gmail.com>
Acked-by: NAlbert Lee <albertcc@tw.ibm.com>
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

The default_disable() function was changed in commit:

 76d21601
 genirq: do not mask interrupts by default

It removed the mask function in favour of the default delayed
interrupt disabling. Unfortunately this also broke the shutdown in
free_irq() when the last handler is removed from the interrupt for
those architectures which rely on the default implementations. Now we
can end up with a enabled interrupt line after the last handler was
removed, which can result in spurious interrupts.

Fix this by adding a default_shutdown function, which is only
installed, when the irqchip implementation does provide neither a
shutdown nor a disable function.

[@stable: affected versions: .21 - .24 ]
Pointed-out-by: NMichael Hennerich <Michael.Hennerich@analog.com>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Acked-by: NIngo Molnar <mingo@elte.hu>
Cc: stable@kernel.org
Tested-by: NMichael Hennerich <Michael.Hennerich@analog.com>

The functions time_before, time_before_eq, time_after, and
time_after_eq are more robust for comparing jiffies against other
values.

So following patch implements usage of the time_after() macro, defined
at linux/jiffies.h, which deals with wrapping correctly
Signed-off-by: NS.Caglar Onur <caglar@pardus.org.tr>
Acked-by: NIngo Molnar <mingo@elte.hu>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>

Currently we fail if someone requests a valid io scheduler, but it's
modular and not currently loaded. That can happen from a driver init
asking for a different scheduler, or online switching through sysfs
as requested by a user.

This patch makes elevator_get() request_module() to attempt to load
the appropriate module, instead of requiring that done manually.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

It's cumbersome to browse a radix tree from start to finish, especially
since we modify keys when a process exits. So add a hlist for the single
purpose of browsing over all known cfq_io_contexts, used for exit,
io prio change, etc.

This fixes http://bugzilla.kernel.org/show_bug.cgi?id=9948Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

That way the interface is symmetric, and calling blk_rq_unmap_user()
on the request wont oops.
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

Commit b2e895db #if 0'ed this code stating:

<--  snip  -->

    [PATCH] revert blockdev direct io back to 2.6.19 version

    Andrew Vasquez is reporting as-iosched oopses and a 65% throughput
    slowdown due to the recent special-casing of direct-io against
    blockdevs.  We don't know why either of these things are occurring.

    The patch minimally reverts us back to the 2.6.19 code for a 2.6.20
    release.

<--  snip  -->

It has since been dead code, and unless someone wants to revive it now
it's time to remove it.

This patch also makes bio_release_pages() static again and removes the
ki_bio_count member from struct kiocb, reverting changes that had been
done for this dead code.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NJens Axboe <axboe@carl.home.kernel.dk>

This patch makes the needlessly global struct def_blk_aops static.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NJens Axboe <axboe@carl.home.kernel.dk>

blk_settings_init() can become static.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NJens Axboe <axboe@carl.home.kernel.dk>

blk_ioc_init() can become static.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NJens Axboe <axboe@carl.home.kernel.dk>

request_cachep needlessly became global.
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NJens Axboe <axboe@carl.home.kernel.dk>

Needed for kretprobes.

Noticed by Ananth N Mavinakayanahalli.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NDavid S. Miller <davem@davemloft.net>