Under certain circumstances lockdep finds an inconsistent lock state in
gadgetfs. The problem can be reproduced with a hardware using the
ci13xxx_udc driver and the gadgetfs test program (needs a patch to support
the ci13xxx_udc, though):
	http://www.linux-usb.org/gadget/usb.c
Start the test program, wait to initialize, then press Ctrl+c.

This patch fixes the following problem by using spin_lock_irqsave()
instead of spin_lock().

=================================
[ INFO: inconsistent lock state ]
3.1.0-rc6+ #158
---------------------------------
inconsistent {IN-HARDIRQ-W} -> {HARDIRQ-ON-W} usage.
usb/113 [HC0[0]:SC0[0]:HE1:SE1] takes:
 (&(&dev->lock)->rlock){?.....}, at: [<bf000340>] gadgetfs_disconnect+0x14/0x80 [gadgetfs]
{IN-HARDIRQ-W} state was registered at:
  [<c00596b8>] mark_irqflags+0x14c/0x1ac
  [<c0059bf8>] __lock_acquire+0x4e0/0x8f0
  [<c005a698>] lock_acquire+0x98/0x1a8
  [<c02f10e0>] _raw_spin_lock+0x54/0x8c
  [<bf000340>] gadgetfs_disconnect+0x14/0x80 [gadgetfs]
  [<c0229104>] _gadget_stop_activity+0xd4/0x154
  [<c022b130>] isr_reset_handler+0x34/0x1c0
  [<c022c320>] udc_irq+0x204/0x228
  [<c0069018>] handle_irq_event_percpu+0x64/0x3a0
  [<c0069390>] handle_irq_event+0x3c/0x5c
  [<c006ae5c>] handle_level_irq+0x8c/0x10c
  [<c0068a34>] generic_handle_irq+0x30/0x44
  [<c0009b2c>] handle_IRQ+0x30/0x84
  [<c0008ef8>] __irq_svc+0x38/0x60
  [<c0009c58>] default_idle+0x30/0x34
  [<c0009e30>] cpu_idle+0x9c/0xd8
  [<c04056f4>] start_kernel+0x278/0x2bc
irq event stamp: 6412
hardirqs last  enabled at (6412): [<c02f1cd0>] _raw_spin_unlock_irqrestore+0x30/0x5c
hardirqs last disabled at (6411): [<c02f1278>] _raw_spin_lock_irqsave+0x20/0xa0
softirqs last  enabled at (6381): [<c002833c>] irq_exit+0xa0/0xa8
softirqs last disabled at (6372): [<c002833c>] irq_exit+0xa0/0xa8

other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(&(&dev->lock)->rlock);
  <Interrupt>
    lock(&(&dev->lock)->rlock);

 *** DEADLOCK ***

1 lock held by usb/113:
 #0:  (udc_lock#2){+.+.+.}, at: [<c02286c0>] usb_gadget_unregister_driver+0x34/0x88

stack backtrace:
[<c000d41c>] (unwind_backtrace+0x0/0xf0) from [<c0058e98>] (print_usage_bug+0x144/0x1c4)
[<c0058e98>] (print_usage_bug+0x144/0x1c4) from [<c0059144>] (mark_lock_irq+0x22c/0x274)
[<c0059144>] (mark_lock_irq+0x22c/0x274) from [<c00592d4>] (mark_lock+0x148/0x3e0)
[<c00592d4>] (mark_lock+0x148/0x3e0) from [<c0059668>] (mark_irqflags+0xfc/0x1ac)
[<c0059668>] (mark_irqflags+0xfc/0x1ac) from [<c0059bf8>] (__lock_acquire+0x4e0/0x8f0)
[<c0059bf8>] (__lock_acquire+0x4e0/0x8f0) from [<c005a698>] (lock_acquire+0x98/0x1a8)
[<c005a698>] (lock_acquire+0x98/0x1a8) from [<c02f10e0>] (_raw_spin_lock+0x54/0x8c)
[<c02f10e0>] (_raw_spin_lock+0x54/0x8c) from [<bf000340>] (gadgetfs_disconnect+0x14/0x80 [gadgetfs])
[<bf000340>] (gadgetfs_disconnect+0x14/0x80 [gadgetfs]) from [<c0229104>] (_gadget_stop_activity+0xd4/0x154)
[<c0229104>] (_gadget_stop_activity+0xd4/0x154) from [<c0229240>] (ci13xxx_stop+0xbc/0x17c)
[<c0229240>] (ci13xxx_stop+0xbc/0x17c) from [<c022867c>] (usb_gadget_remove_driver+0x88/0x98)
[<c022867c>] (usb_gadget_remove_driver+0x88/0x98) from [<c02286f4>] (usb_gadget_unregister_driver+0x68/0x88)
[<c02286f4>] (usb_gadget_unregister_driver+0x68/0x88) from [<bf0003e8>] (dev_release+0x14/0x48 [gadgetfs])
[<bf0003e8>] (dev_release+0x14/0x48 [gadgetfs]) from [<c00cc158>] (__fput+0xa4/0x1f0)
[<c00cc158>] (__fput+0xa4/0x1f0) from [<c00c7f28>] (filp_close+0x5c/0x74)
[<c00c7f28>] (filp_close+0x5c/0x74) from [<c00c7fe8>] (sys_close+0xa8/0x150)
[<c00c7fe8>] (sys_close+0xa8/0x150) from [<c00092a0>] (ret_fast_syscall+0x0/0x38)
Tested-by: NPavankumar Kondeti <pkondeti@codeaurora.org>
Signed-off-by: NMarc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: NFelipe Balbi <balbi@ti.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

remove the following two paragraphs as they are not needed:

This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
License for more details.

You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc.,59
Temple Place - Suite 330, Boston, MA  02111-1307, USA.
Signed-off-by: NKlaus Schwarzkopf <schwarzkopf@sensortherm.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

maxpacket is set by the udc driver for ep0 very early. This value is
copied by the function gadget used later for the USB_DT_DEVICE and
USB_DT_DEVICE_QUALIFIER query. This seems to work fine so far. For USB3
we need set a different value here. In SS speed it is 2^x with x=9 and
in HS we set something <= 64. If the UDC starts in SS and continues in
HS after the cable has been plugged it will report a too small value.
There setting of this value is defered and taken automaticly from the
ep0 pointer where the UDC driver can update it according to the speed it
detected _after_ a cable has been plugged.
Signed-off-by: NSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: NFelipe Balbi <balbi@ti.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Change usb_ep_enable() prototype to use endpoint
descriptor from usb_ep.

This optimization spares the FDs from saving the
endpoint chosen descriptor. This optimization is
not full though. To fully exploit this change, one
needs to update all the UDCs as well since in the
current implementation each of them saves the
endpoint descriptor in it's internal (and extended)
endpoint structure.
Signed-off-by: NTatyana Brokhman <tlinder@codeaurora.org>
Signed-off-by: NFelipe Balbi <balbi@ti.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

ep_write() acquires data->lock mutex in get_ready_ep() and releases it
on all paths except for one: when usb_endpoint_xfer_isoc() failed. The
patch adds mutex_unlock(&data->lock) at that path.

It is similar to commit 00cc7a5f ("usb-gadget: unlock data->lock mutex on error path in ep_read()"),
it was not fixed at that time by accident.

Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: NAlexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

ep_read() acquires data->lock mutex in get_ready_ep() and releases it on
all paths except for one: when usb_endpoint_xfer_isoc() failed. The
patch adds mutex_unlock(&data->lock) at that path.

Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: NAlexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

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

Instead of always assigning an increasing inode number in new_inode
move the call to assign it into those callers that actually need it.
For now callers that need it is estimated conservatively, that is
the call is added to all filesystems that do not assign an i_ino
by themselves.  For a few more filesystems we can avoid assigning
any inode number given that they aren't user visible, and for others
it could be done lazily when an inode number is actually needed,
but that's left for later patches.
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NDave Chinner <dchinner@redhat.com>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

To accomplish this the function to register a gadget driver takes the bind
function as a second argument.  To make things clearer rename the function
to resemble platform_driver_probe.

This fixes many section mismatches like

	WARNING: drivers/usb/gadget/g_printer.o(.data+0xc): Section mismatch in
	reference from the variable printer_driver to the function
	.init.text:printer_bind()
	The variable printer_driver references
	the function __init printer_bind()

All callers are fixed.
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
[m.nazarewicz@samsung.com: added dbgp]
Signed-off-by: NMichał Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The BKL is only used in fill_super, which is  protected by the superblocks
s_umount rw_semaphore. Therefore it is safe to remove the BKL entirely.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: linux-usb@vger.kernel.org

There is no gadget driver in the tree that
actually implements the ioctl operation, so
obviously it is not necessary to hold the
BKL around the call.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Michał Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Use memdup_user when user data is immediately copied into the
allocated region.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression from,to,size,flag;
position p;
identifier l1,l2;
@@

-  to = \(kmalloc@p\|kzalloc@p\)(size,flag);
+  to = memdup_user(from,size);
   if (
-      to==NULL
+      IS_ERR(to)
                 || ...) {
   <+... when != goto l1;
-  -ENOMEM
+  PTR_ERR(to)
   ...+>
   }
-  if (copy_from_user(to, from, size) != 0) {
-    <+... when != goto l2;
-    -EFAULT
-    ...+>
-  }
// </smpl>
Signed-off-by: NJulia Lawall <julia@diku.dk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The semaphore data->lock is semantically a mutex. Convert it to a real
mutex.
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Use helper functions to determine the type and direction of an endpoint
instead of fiddling with bEndpointAddress and bmAttributes
Signed-off-by: NMatthias Kaehlcke <matthias@kaehlcke.net>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The base versions handle constant folding now.
Signed-off-by: NHarvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

... and don't bother in callers.  Don't bother with zeroing i_blocks,
while we are at it - it's already been zeroed.

i_mode is not worth the effort; it has no common default value.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

As it is, all instances of ->release() for files that have ->fasync()
need to remember to evict file from fasync lists; forgetting that
creates a hole and we actually have a bunch that *does* forget.

So let's keep our lives simple - let __fput() check FASYNC in
file->f_flags and call ->fasync() there if it's been set.  And lose that
crap in ->release() instances - leaving it there is still valid, but we
don't have to bother anymore.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

We want to use WARN() as a variant of WARN_ON(), however a few drivers are
using WARN() internally.  This patch renames these to WARNING() to avoid the
namespace clash.  A few cases were defining but not using the thing, for those
cases I just deleted the definition.
Signed-off-by: NArjan van de Ven <arjan@linux.intel.com>
Acked-by: NGreg KH <greg@kroah.com>
Cc: Karsten Keil <kkeil@suse.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This keeps the gadget ioctl method wrapped but pushes the BKL down into
the gadget code so we can use unlocked_ioctl().
Signed-off-by: NAlan Cox <alan@redhat.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The pxa2xx_udc.c driver is renamed to pxa25x_udc.c (the platform
driver name changes from pxa2xx-udc to pxa25x-udc) and the
platform driver name of pxa27x_udc.c is fixed to pxa27x-udc.
pxa_device_udc in devices.c is split into pxa25x and pxa27x flavors
and the pxa27x_device_udc is enabled in pxa27x.c.
Signed-off-by: NPhilipp Zabel <philipp.zabel@gmail.com>
Acked-by: NNicolas Pitre <nico@cam.org>
Acked-by: NEric Miao <eric.miao@marvell.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Including from Ian Molton:

Fixes for mistakes left over from the PXA2{5,7}X UDC split.
Signed-off-by: NIan Molton <spyro@f2s.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

__FUNCTION__ is gcc-specific, use __func__
Signed-off-by: NHarvey Harrison <harvey.harrison@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

gadgetfs (drivers/usb/gadget/inode.c) was not delegating all
non-device requests to userspace.  This patch makes the handling of
all request cases consistent.
Signed-off-by: NRoy Hashimoto <hashimot@alumni.caltech.edu>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

We now have pr_err(), pr_warning(), and friends ... start using
them in the gadget stack instead of printk(KERN_ERR) and friends.
This gives us shorter lines and somewhat increased readability.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Move <linux/usb_gadget.h> to <linux/usb/gadget.h>, reducing
some of the clutter in the main include directory.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This adds two small inlines to the gadget stack, which will
often evaluate to compile-time constants.  That can help
shrink object code and remove #ifdeffery.

 - gadget_is_dualspeed(), currently always a compile-time
   constant (depending on which controller is selected).

 - gadget_is_otg(), usually a compile time "false", but this
   is a runtime test if the platform enables OTG (since it's
   reasonable to populate boards with different USB sockets).

It also updates two peripheral controller drivers to use these:

 - fsl_usb2_udc, mostly OTG-related bugfixes:  non-OTG devices
   must follow the rules about drawing VBUS power, and OTG ones
   need to reject invalid SET_FEATURE requests.

 - omap_udc, just scrubbing a bit of #ifdeffery.

And also gadgetfs, which lost some #ifdefs and moved to a more
standard handling of DEBUG and VERBOSE_DEBUG.

The main benefits come from patches which will follow.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This removes complaints about the gadget stack which are generated by
the currrent "sparse":  it doesn't like the fact that zero is the null
pointer.  (Last I checked, C guarantees that's correct ...)
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Remove usb_ep_{alloc,free}_buffer() calls, for small dma-coherent buffers.
This patch just removes the interface and its users; later patches will
remove controller driver support.

  - This interface is invariably not implemented correctly in the
    controller drivers (e.g. using dma pools, a mechanism which
    post-dates the interface by several years).

  - At this point no gadget driver really *needs* to use it.  In
    current kernels, any driver that needs such a mechanism could
    allocate a dma pool themselves.

Removing this interface is thus a simplification and improvement.

Note that the gmidi.c driver had a bug in this area; fixed.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Make sure gadgetfs userspace interface is properly exported:

 - Move <linux/usb_gadgetfs.h> to <linux/usb/gadgetfs.h>;
 - Export it using Kbuild;
 - Add an #include guard;
 - Correct some internal documentation;
 - Update struct layout so it's the same on 32/64 bit kernels.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Remove some dead CONFIG_ symbols, and document the status of a few others.
The "gadget_chips.h" references are by and large to drivers which exist
but haven't yet been submitted for merging to the main 2.6 tree.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

It turns out that le16_to_cpup() and le32_to_cpup() aren't always safe
to call with pointers into packed structures, since those are inlined
functions and GCC may lose the "packed" attribute.  So those references
can become unaligned kernel accesses, which are evil on some hardware.

This patch updates uses of those routines in the gadget stack.  The
references into packed structures can just use leXX_to_cpu(*x), which
in most cases is more natural.  Some other uses in RNDIS, mostly in
debug code, were wrong in the first place; those use get_unaligned().
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

I don't think the current code works with multiple iovecs.
The original would just copy the first part of priv->buf
over and over into multiple iovecs.
Signed-off-by: NSarah Bailey <saharabeara@gmail.com>
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch (as837) fixes several mistakes in the AIO interface of the
gadgetfs driver:

	The ki_retry method is not supposed to do a put on the kiocb.
	The extra call to aio_put_req() causes memory corruption.
	(Note: This call was removed before, by patch as691, and then
	mysteriously re-introduced later.)

	Even if a read transfer is cancelled, we can and should send
	to the user all the data that did manage to get transferred.

	Testing for AIO cancellation in the I/O completion handler
	is both racy and (now) unnecessary.  aio_complete() does its
	own checking, in a safe manner.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Resolve an initizlization issue that could come up if the userspace
driver wrote invalid descriptors to a dual-speed device.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This resolves a race in gadgetfs associated with changing device/ep0
when processing control requests.  The fix is to change that state
earlier, when the control response is issued, so there's no window
in which userspace could see the wrong state; and enlarge the scope
of the spinlock during the ep0 request completion handler.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This simplifies event reading by eliminating arithmetic and being
more direct/obvious, and tweaks some debug messages slightly.
The math elimination will change timings, sometimes enough to
allow a race to appear.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Minor gadgetfs cleanups:

 - EP0 state constants become consistently STATE_DEV_* rather than
   sometimes omitting the "DEV_"; STATE_EP_* were already consistent.

 - Comment that ep0 state is protected by the spinlock, and update
   code that was neglecting that rule.

None of this is expected to change behavior.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Gadgetfs had a mode in which endpoint descriptors were written by the user
program before connection.  This mode had some bugs, and hasn't seen much
(if any) use.  This patch removes that mode, leaving the mode of operation
where the user program waits for endpoint 0 to report a SET_CONFIGURATION,
and only then configures the endpoints.

From: "Phil Endecott" <spam_from_usb_devel@chezphil.org>
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Remove some whitespace bugs in gadgetfs (mostly from someone's
patch updating the AIO support).
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>