Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

no more users left outside of fs/*.c (and very few outside of
fs/namespace.c, actually)
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

apply function to vfsmounts in set returned by collect_mounts(),
stop if it returns non-zero.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Analog of is_subdir for vfsmount,dentry pairs, moved from audit_tree.c
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

The handling of mount flags in set_mnt_shared() got a little tangled
up during previous cleanups, with the following problems:

* MNT_PNODE_MASK is defined as a literal constant when it should be a
bitwise xor of other MNT_* flags
* set_mnt_shared() clears and then sets MNT_SHARED (part of MNT_PNODE_MASK)
* MNT_PNODE_MASK could use a comment in mount.h
* MNT_PNODE_MASK is a terrible name, change to MNT_SHARED_MASK

This patch fixes these problems.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

No one is calling this anymore as everyone has switched to
invalidate_mapping_pages long time ago.  Also update a few
references to it in comments.  nfs has two more, but I can't
easily figure what they are actually referring to, so I left
them as-is.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

re-order structure super_block to remove 16 bytes of alignment padding
on 64bit builds.

This shrinks the size of super_block from 712 to 696 bytes so requiring
one fewer 64 byte cache lines.
Signed-off-by: NRichard Kennedy <richard@rsk.demon.co.uk>

-----
patch against 2.6.33-rc5
compiled & tested on x86_64 AMDX2 desktop machine.

I've been running with this patch applied for several weeks with no
problems.

regards
Richard
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Remove the EXPORT_UNUSED_SYMBOL of simple_prepare_write

Collapse simple_prepare_write into it's only caller, though
making it simpler and clearer to understand.
Signed-off-by: NBoaz Harrosh <bharrosh@panasas.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

This reverts commit fb1e7538.

"Benjamin S." <sbenni@gmx.de> reports that the patch in question
causes a big drop in sequential throughput for him, dropping from
200MB/sec down to only 70MB/sec.

Needs to be investigated more fully, for now lets just revert the
offending commit.

Conflicts:

	include/linux/blkdev.h
Signed-off-by: NJens Axboe <jens.axboe@oracle.com>

This fixes the filepath encoded in <linux/amba/bus.h> and adds
some documentation as to what this bus really means.
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

This removes 8 bytes of padding from struct inode on 64bit builds, and
so allows 1 more object/slab in the inode_cache when using slub.
Signed-off-by: NRichard Kennedy <richard@rsk.demon.co.uk>
----
patch against 2.6.33-rc8
compiled & tested on x86_64 AMDX2

I've been running this patch for over a week with no obvious problems
regards
Richard
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Most laptops have keys that are intended to toggle all device state, not
just wifi. These are currently generally mapped to KEY_WLAN. As a result,
rfkill will only kill or enable wifi in response to the key press. This
confuses users and can make it difficult for them to enable bluetooth
and wwan devices.

This patch adds a new keycode, KEY_RFKILL. It indicates that the system
should toggle the state of all rfkillable devices.
Signed-off-by: NMatthew Garrett <mjg@redhat.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NDmitry Torokhov <dtor@mail.ru>

This patch fixes following sparse warnings:

include/linux/kfifo.h:127:25: warning: Using plain integer as NULL pointer
kernel/kfifo.c:83:21: warning: Using plain integer as NULL pointer
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: NStefani Seibold <stefani@seibold.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Even if this bumps the version to 1 it does not mean the driver is
out of staging. From what we know this is the last backwards
incompatible change to the driver.
Signed-off-by: NJakob Bornecrantz <jakob@vmware.com>
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

When time-based throttling is implemented, we need to bump minor.
When the old way of detecting scanout is removed, we need to bump major.
In the meantime, this change should not break existing user-space.
Signed-off-by: NThomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: NJakob Bornecrantz <jakob@vmware.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

On nv50, this will be needed by applications using CUDA to know
how much stack/local memory to allocate.
Signed-off-by: NMarcin Kościelnicki <koriakin@0x04.net>
Signed-off-by: NBen Skeggs <bskeggs@redhat.com>

As noticed by Jon Masters <jonathan@jonmasters.org>, the conntrack hash
size is global and not per namespace, but modifiable at runtime through
/sys/module/nf_conntrack/hashsize. Changing the hash size will only
resize the hash in the current namespace however, so other namespaces
will use an invalid hash size. This can cause crashes when enlarging
the hashsize, or false negative lookups when shrinking it.

Move the hash size into the per-namespace data and only use the global
hash size to initialize the per-namespace value when instanciating a
new namespace. Additionally restrict hash resizing to init_net for
now as other namespaces are not handled currently.

Cc: stable@kernel.org
Signed-off-by: NPatrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

nf_conntrack_cachep is currently shared by all netns instances, but
because of SLAB_DESTROY_BY_RCU special semantics, this is wrong.

If we use a shared slab cache, one object can instantly flight between
one hash table (netns ONE) to another one (netns TWO), and concurrent
reader (doing a lookup in netns ONE, 'finding' an object of netns TWO)
can be fooled without notice, because no RCU grace period has to be
observed between object freeing and its reuse.

We dont have this problem with UDP/TCP slab caches because TCP/UDP
hashtables are global to the machine (and each object has a pointer to
its netns).

If we use per netns conntrack hash tables, we also *must* use per netns
conntrack slab caches, to guarantee an object can not escape from one
namespace to another one.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
[Patrick: added unique slab name allocation]
Cc: stable@kernel.org
Signed-off-by: NPatrick McHardy <kaber@trash.net>

ima_path_check actually deals with files!  call it ima_file_check instead.
Signed-off-by: NEric Paris <eparis@redhat.com>
Acked-by: NMimi Zohar <zohar@linux.vnet.ibm.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

The "Untangling ima mess, part 2 with counters" patch messed
up the counters.  Based on conversations with Al Viro, this patch
streamlines ima_path_check() by removing the counter maintaince.
The counters are now updated independently, from measuring the file,
in __dentry_open() and alloc_file() by calling ima_counts_get().
ima_path_check() is called from nfsd and do_filp_open().
It also did not measure all files that should have been measured.
Reason: ima_path_check() got bogus value passed as mask.
[AV: mea culpa]
[AV: add missing nfsd bits]
Signed-off-by: NMimi Zohar <zohar@us.ibm.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

This is to make the annotation of percpu variables during the next merge
window less painfull.

Extracted from a patch by Rusty Russell.
Signed-off-by: NStephen Rothwell <sfr@canb.auug.org.au>
Acked-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

The value we get from the low byte of the ATA_ID_SECTOR_SIZE word is not not
a plain multiple, but the log of it, so fix the helper to give the correct
answer.  Without this we'll get an incorrect minimal I/O size in the block
limits VPD page for 4k sector drives.

Also change the return value of ata_id_logical_per_physical_sectors to u16
for the unlikely case of very large logical sectors.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NJeff Garzik <jgarzik@redhat.com>

Change 'bp_len' type to __u64 to make it work across archs as
the s390 architecture watch point length can be upto 2^64.

reference:
	http://lkml.org/lkml/2010/1/25/212

This is an ABI change that is not backward compatible with
the previous hardware breakpoint info layout integrated in this
development cycle, a rebuilt of perf tools is necessary for
versions based on 2.6.33-rc1 - 2.6.33-rc6 to work with a
kernel based on this patch.
Signed-off-by: NMahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Acked-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: "K. Prasad" <prasad@linux.vnet.ibm.com>
Cc: Maneesh Soni <maneesh@in.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Martin <schwidefsky@de.ibm.com>
LKML-Reference: <20100130045518.GA20776@in.ibm.com>
Signed-off-by: NFrederic Weisbecker <fweisbec@gmail.com>

On Tue, Feb 02, 2010 at 02:57:14PM -0800, Greg KH (gregkh@suse.de) wrote:
> > There are at least two ways to fix it: using a big cannon and a small
> > one. The former way is to disable notification registration, since it is
> > not used by anyone at all. Second way is to check whether calling
> > process is root and its destination group is -1 (kind of priveledged
> > one) before command is dispatched to workqueue.
> 
> Well if no one is using it, removing it makes the most sense, right?
> 
> No objection from me, care to make up a patch either way for this?

Getting it is not used, let's drop support for notifications about
(un)registered events from connector.
Another option was to check credentials on receiving, but we can always
restore it without bugs if needed, but genetlink has a wider code base
and none complained, that userspace can not get notification when some
other clients were (un)registered.

Kudos for Sebastian Krahmer <krahmer@suse.de>, who found a bug in the
code.
Signed-off-by: NEvgeniy Polyakov <zbr@ioremap.net>
Acked-by: NGreg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When CONFIG_HAVE_UNSTABLE_SCHED_CLOCK is set, sched_clock() gets
the time from hardware such as the TSC on x86. In this
configuration kgdb will report a softlock warning message on
resuming or detaching from a debug session.

Sequence of events in the problem case:

 1) "cpu sched clock" and "hardware time" are at 100 sec prior
    to a call to kgdb_handle_exception()

 2) Debugger waits in kgdb_handle_exception() for 80 sec and on
    exit the following is called ...  touch_softlockup_watchdog() -->
    __raw_get_cpu_var(touch_timestamp) = 0;

 3) "cpu sched clock" = 100s (it was not updated, because the
    interrupt was disabled in kgdb) but the "hardware time" = 180 sec

 4) The first timer interrupt after resuming from
    kgdb_handle_exception updates the watchdog from the "cpu sched clock"

update_process_times() { ...  run_local_timers() -->
softlockup_tick() --> check (touch_timestamp == 0) (it is "YES"
here, we have set "touch_timestamp = 0" at kgdb) -->
__touch_softlockup_watchdog() ***(A)--> reset "touch_timestamp"
to "get_timestamp()" (Here, the "touch_timestamp" will still be
set to 100s.)  ...

    scheduler_tick() ***(B)--> sched_clock_tick() (update "cpu sched
    clock" to "hardware time" = 180s) ...  }

 5) The Second timer interrupt handler appears to have a large
    jump and trips the softlockup warning.

update_process_times() { ...  run_local_timers() -->
softlockup_tick() --> "cpu sched clock" - "touch_timestamp" =
180s-100s > 60s --> printk "soft lockup error messages" ...  }

note: ***(A) reset "touch_timestamp" to
"get_timestamp(this_cpu)"

Why is "touch_timestamp" 100 sec, instead of 180 sec?

When CONFIG_HAVE_UNSTABLE_SCHED_CLOCK is set, the call trace of
get_timestamp() is:

get_timestamp(this_cpu)
 -->cpu_clock(this_cpu)
 -->sched_clock_cpu(this_cpu)
 -->__update_sched_clock(sched_clock_data, now)

The __update_sched_clock() function uses the GTOD tick value to
create a window to normalize the "now" values.  So if "now"
value is too big for sched_clock_data, it will be ignored.

The fix is to invoke sched_clock_tick() to update "cpu sched
clock" in order to recover from this state.  This is done by
introducing the function touch_softlockup_watchdog_sync(). This
allows kgdb to request that the sched clock is updated when the
watchdog thread runs the first time after a resume from kgdb.

[yong.zhang0@gmail.com: Use per cpu instead of an array]
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
Signed-off-by: NDongdong Deng <Dongdong.Deng@windriver.com>
Cc: kgdb-bugreport@lists.sourceforge.net
Cc: peterz@infradead.org
LKML-Reference: <1264631124-4837-2-git-send-email-jason.wessel@windriver.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Signed-off-by: NDave Airlie <airlied@redhat.com>

This patch fixes the regression in functionality where the
kernel debugger and the perf API do not nicely share hw
breakpoint reservations.

The kernel debugger cannot use any mutex_lock() calls because it
can start the kernel running from an invalid context.

A mutex free version of the reservation API needed to get
created for the kernel debugger to safely update hw breakpoint
reservations.

The possibility for a breakpoint reservation to be concurrently
processed at the time that kgdb interrupts the system is
improbable. Should this corner case occur the end user is
warned, and the kernel debugger will prohibit updating the
hardware breakpoint reservations.

Any time the kernel debugger reserves a hardware breakpoint it
will be a system wide reservation.
Signed-off-by: NJason Wessel <jason.wessel@windriver.com>
Acked-by: NFrederic Weisbecker <fweisbec@gmail.com>
Cc: kgdb-bugreport@lists.sourceforge.net
Cc: K.Prasad <prasad@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: torvalds@linux-foundation.org
LKML-Reference: <1264719883-7285-3-git-send-email-jason.wessel@windriver.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

'flush_old_exec()' is the point of no return when doing an execve(), and
it is pretty badly misnamed.  It doesn't just flush the old executable
environment, it also starts up the new one.

Which is very inconvenient for things like setting up the new
personality, because we want the new personality to affect the starting
of the new environment, but at the same time we do _not_ want the new
personality to take effect if flushing the old one fails.

As a result, the x86-64 '32-bit' personality is actually done using this
insane "I'm going to change the ABI, but I haven't done it yet" bit
(TIF_ABI_PENDING), with SET_PERSONALITY() not actually setting the
personality, but just the "pending" bit, so that "flush_thread()" can do
the actual personality magic.

This patch in no way changes any of that insanity, but it does split the
'flush_old_exec()' function up into a preparatory part that can fail
(still called flush_old_exec()), and a new part that will actually set
up the new exec environment (setup_new_exec()).  All callers are changed
to trivially comply with the new world order.
Signed-off-by: NH. Peter Anvin <hpa@zytor.com>
Cc: stable@kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

For pressure-based multi-touch devices, a direct way to send sensor
intensity data per finger is needed. This patch adds the ABS_MT_PRESSURE
event to the MT protocol.
Requested-by: NYoonyoung Shim <jy0922.shim@samsung.com>
Requested-by: NMika Kuoppala <mika.kuoppala@nokia.com>
Requested-by: NPeter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: NHenrik Rydberg <rydberg@euromail.se>
Signed-off-by: NDmitry Torokhov <dtor@mail.ru>

We are missing these when building the pci_dev from scratch off
the Open Firmware device-tree
Signed-off-by: NBenjamin Herrenschmidt <benh@kernel.crashing.org>
Acked-by: NJesse Barnes <jbarnes@virtuousgeek.org>

It's a simplified 'read_cache_page()' which takes a page allocation
flag, so that different paths can control how aggressive the memory
allocations are that populate a address space.

In particular, the intel GPU object mapping code wants to be able to do
a certain amount of own internal memory management by automatically
shrinking the address space when memory starts getting tight.  This
allows it to dynamically use different memory allocation policies on a
per-allocation basis, rather than depend on the (static) address space
gfp policy.

The actual new function is a one-liner, but re-organizing the helper
functions to the point where you can do this with a single line of code
is what most of the patch is all about.
Tested-by: NChris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

struct fw_cdev_add_descriptor.length is in quadlets, not in bytes.
Also remove any doubts about the endianess of descriptor data.
Signed-off-by: NStefan Richter <stefanr@s5r6.in-berlin.de>

KVM needs a wait to atomically remove themselves from the eventfd ->poll()
wait queue head, in order to handle correctly their IRQfd deassign
operation.

This patch introduces such API, plus a way to read an eventfd from its
context.
Signed-off-by: NDavide Libenzi <davidel@xmailserver.org>
Signed-off-by: NAvi Kivity <avi@redhat.com>

GC is non-existent in netns, so after you hit GC threshold, no new
dst entries will be created until someone triggers cleanup in init_net.

Make xfrm4_dst_ops and xfrm6_dst_ops per-netns.
This is not done in a generic way, because it woule waste
(AF_MAX - 2) * sizeof(struct dst_ops) bytes per-netns.

Reorder GC threshold initialization so it'd be done before registering
XFRM policies.
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

"ip xfrm state|policy count" report SA/SP count from init_net,
not from netns of caller process.
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The is_software_event() definition always confuses me because its an
exclusive expression, make it an inclusive one.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

SD_PREFER_SIBLING is set at the CPU domain level if power saving isn't
enabled, leading to many cache misses on large machines as we traverse
looking for an idle shared cache to wake to.  Change the enabler of
select_idle_sibling() to SD_SHARE_PKG_RESOURCES, and enable same at the
sibling domain level.
Reported-by: NLin Ming <ming.m.lin@intel.com>
Signed-off-by: NMike Galbraith <efault@gmx.de>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1262612696.15495.15.camel@marge.simson.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Borislav Petkov reports issues with duplicate sysfs endpoint files after a
resume from a hibernate.  It turns out that the code to support alternate
settings under xHCI has issues when a device with a non-default alternate
setting is reset during the hibernate:

[  427.681810] Restarting tasks ...
[  427.681995] hub 1-0:1.0: state 7 ports 6 chg 0004 evt 0000
[  427.682019] usb usb3: usb resume
[  427.682030] ohci_hcd 0000:00:12.0: wakeup root hub
[  427.682191] hub 1-0:1.0: port 2, status 0501, change 0000, 480 Mb/s
[  427.682205] usb 1-2: usb wakeup-resume
[  427.682226] usb 1-2: finish reset-resume
[  427.682886] done.
[  427.734658] ehci_hcd 0000:00:12.2: port 2 high speed
[  427.734663] ehci_hcd 0000:00:12.2: GetStatus port 2 status 001005 POWER sig=se0 PE CONNECT
[  427.746682] hub 3-0:1.0: hub_reset_resume
[  427.746693] hub 3-0:1.0: trying to enable port power on non-switchable hub
[  427.786715] usb 1-2: reset high speed USB device using ehci_hcd and address 2
[  427.839653] ehci_hcd 0000:00:12.2: port 2 high speed
[  427.839666] ehci_hcd 0000:00:12.2: GetStatus port 2 status 001005 POWER sig=se0 PE CONNECT
[  427.847717] ohci_hcd 0000:00:12.0: GetStatus roothub.portstatus [1] = 0x00010100 CSC PPS
[  427.915497] hub 1-2:1.0: remove_intf_ep_devs: if: ffff88022f9e8800 ->ep_devs_created: 1
[  427.915774] hub 1-2:1.0: remove_intf_ep_devs: bNumEndpoints: 1
[  427.915934] hub 1-2:1.0: if: ffff88022f9e8800: endpoint devs removed.
[  427.916158] hub 1-2:1.0: create_intf_ep_devs: if: ffff88022f9e8800 ->ep_devs_created: 0, ->unregistering: 0
[  427.916434] hub 1-2:1.0: create_intf_ep_devs: bNumEndpoints: 1
[  427.916609]  ep_81: create, parent hub
[  427.916632] ------------[ cut here ]------------
[  427.916644] WARNING: at fs/sysfs/dir.c:477 sysfs_add_one+0x82/0x96()
[  427.916649] Hardware name: System Product Name
[  427.916653] sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:12.2/usb1/1-2/1-2:1.0/ep_81'
[  427.916658] Modules linked in: binfmt_misc kvm_amd kvm powernow_k8 cpufreq_ondemand cpufreq_powersave cpufreq_userspace freq_table cpufreq_conservative ipv6 vfat fat
+8250_pnp 8250 pcspkr ohci_hcd serial_core k10temp edac_core
[  427.916694] Pid: 278, comm: khubd Not tainted 2.6.33-rc2-00187-g08d869aa-dirty #13
[  427.916699] Call Trace:

The problem is caused by a mismatch between the USB core's view of the
device state and the USB device and xHCI host's view of the device state.

After the device reset and re-configuration, the device and the xHCI host
think they are using alternate setting 0 of all interfaces.  However, the
USB core keeps track of the old state, which may include non-zero
alternate settings.  It uses intf->cur_altsetting to keep the endpoint
sysfs files for the old state across the reset.

The bandwidth allocation functions need to know what the xHCI host thinks
the current alternate settings are, so original patch set
intf->cur_altsetting to the alternate setting 0.  This caused duplicate
endpoint files to be created.

The solution is to not set intf->cur_altsetting before calling
usb_set_interface() in usb_reset_and_verify_device().  Instead, we add a
new flag to struct usb_interface to tell usb_hcd_alloc_bandwidth() to use
alternate setting 0 as the currently installed alternate setting.
Signed-off-by: NSarah Sharp <sarah.a.sharp@linux.intel.com>
Tested-by: NBorislav Petkov <petkovbb@googlemail.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

commit 541cd3ee ("phylib: Fix deadlock
on resume") caused TI DaVinci EMAC ethernet driver to oops upon resume:

 PM: resume of devices complete after 237.098 msecs
 Restarting tasks ... done.
 kernel BUG at kernel/workqueue.c:354!
 Unable to handle kernel NULL pointer dereference at virtual address 00000000
 [...]
 Backtrace:
 [<c002c598>] (__bug+0x0/0x2c) from [<c0052a54>] (queue_delayed_work_on+0x74/0xf8)
 [<c00529e0>] (queue_delayed_work_on+0x0/0xf8) from [<c0052b30>] (queue_delayed_work+0x2c/0x30)

The oops pops up because TI DaVinci EMAC driver detaches PHY on
suspend and attaches it back on resume. Attaching makes phylib call
phy_start_machine() that initializes a workqueue. On the other hand,
PHY's resume routine will call phy_start_machine() again, and that
will cause the oops since we just destroyed the already scheduled
workqueue.

This patch fixes the issue by moving workqueue initialization to
phy_device_create().

p.s. We don't see this oops with ucc_geth and gianfar drivers because
they perform a fine-grained suspend, i.e. they just stop the PHYs
without detaching.
Reported-by: NSekhar Nori <nsekhar@ti.com>
Signed-off-by: NAnton Vorontsov <avorontsov@ru.mvista.com>
Tested-by: NSekhar Nori <nsekhar@ti.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>