"Static" buffers in fsg_buffhd structure (ie. fields which are arrays
rather then pointers to dynamically allocated memory) are not aligned
to any "big" power of two which may lead to poor DMA performance
(copying "by hand" of head or tail) or no DMA at all even if otherwise
hardware supports it.

Therefore, this patch makes mass storage function use kmalloc()ed
buffers which are (because of their size) page aligned (which should
be enough for any hardware).
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Contrary to the comment in fsg_add, fsg_bind calls fsg_unbind on errors,
which decreases refcount and frees the fsg_dev structure, causing trouble
when fsg_add does the same.

Fix it by simply leaving up cleanup to fsg_add().
Signed-off-by: NPeter Korsgaard <jacmet@sunsite.dk>
Acked-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The recent commit (0e530b45) that moved usb_ep_autoconfig from the
__devinit section to the __init section missed the mass storage device.
Its fsg_bind() function uses the usb_ep_autoconfig() function from non
__init context leading to:
WARNING: drivers/usb/gadget/g_mass_storage.o(.text): Section mismatch in
	reference from the function _fsg_bind()
	            to the function .init.text:_usb_ep_autoconfig()

So move fsg_bind() into __init as well.
Signed-off-by: NMike Frysinger <vapier@gentoo.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

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>

A bunch of USB gadget drivers where never ported from the linux 2.4
series to 2.6 kernels. However there's some code still in the tree for
them which isn't used and is probably untested for ages.

As the chance of these drivers being forward ported is probably quite
small now it might be time to get rid of them.
Signed-off-by: NChristoph Egger <siccegge@stud.informatik.uni-erlangen.de>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

init_completion() hasn't been called yet and the thread isn't created
if we end up here, so don't call complete() on thread_notifier.
Signed-off-by: NPeter Korsgaard <jacmet@sunsite.dk>
Acked-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: stable <stable@kernel.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch fixes warning caused by calling min() macro
with arguments of different types:

  drivers/usb/gadget/f_mass_storage.c:623: warning: \
  comparison of distinct pointer types lacks a cast
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: Marek Szyprowski <m.nazarewicz@samsung.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The mass storage function responded needlessly to a set
configuration packet.  This was a leftover from converting
gadget (file storage gadget) into a composite function.

Moreover, it has failed to respond to get max LUN request.
Adding request queueing made the function work better.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NMarek Szyprowski <m.szyprowski@samsung.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Adds a fallback which forces all LUNs ejection (including
non-removable and with prevent_medium_removal flag) when mass storage
function (MSF) worker thread exits and gadget fails to handle the
situation.

Previously, if thread_exits was not specified mass storage function
(MSF) did nothing when exiting thread as it's unclear for *function*
what to do when it's thread terminates so responsibility of handling
this situation was left to the *gadget* using the function.

The g_mass_storage handled the situation by unregistering itself (the
same thing that file storage gadget does).  However, g_multi did
nothing and so MSF did not eject LUNs which prevented file system
unmounting.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Reviewed-by: NKyungmin Park <kyungmin.park@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

thread_exits callback has been added to fsg_common structure.
This callback is called when MSF's thread exits (is terminated
by a signal or function is unregistered).  It's then gadget's
responsibility to unregister the gadget.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Most of the data from fsg_dev have been moved to fsg_common
structure.  The fsg_dev structure holds only endpoint dependent
data.  The fsg_common structure has a fsg pointer which points
to active fsg_dev structure -- endpoints are referenced via this
pointer.

This fixes the problem of several threads created when a single
instance of MSF is used in several USB configurations.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Fixed most of the errors and warnings in f_mass_storage.c and
storage_common.c reported by checkpatch.pl as well as updated
comments.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

A two fsg_config fields were added:
* lun_name_format which lets one specify format of a name
  used when registering LUN devices.  It is useful if there
  would be ever need for two MSFs to be used in a single
  composite gadget (as opposed to single MSF in two
  configuration); and
* thread_name which lets one specify the name of a kernel
  thread used by MSF.  This is not required since two or more
  threads can have the same name but nevertheless it's here
  for consistency.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Removed all references to mod_data in f_mass_storage.c and
instead created fsg_config structure fsg_common_init() takes
as an argument -- it stores all configuration options that
were previously taken from mod_data.

Moreover, The fsg_config structure allows per-LUN
configuration of removable and CD-ROM emulation.

Module parameters are handled by defining an object of
fsg_module_parameters structure and then declaring module
parameters via FSG_MODULE_PARAMETERS() macro.  It adds proper
declarations to the code making specified object be populated
from module parameters.

To use values stored there one may use either
fsg_config_from_params() which will will a fsg_config structure
with values taken from fsg_module_parameters structure or
fsg_common_from_params() which will initialise fsg_common
structure directly.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The f_mass_storage.c has been changed into a composite function.
mass_storage.c file has been introduced which defines a
g_mass_storage gadget based on composite framework.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Moved code initialising fsg_common structure to fsg_common_init()
function which is called from fsg_bind().  Moreover, changed
reference counting mechanism: fsg_common has a reference counter
which counts how many fsg_dev structures uses it.  When this
reaches zero fsg_common_release() is run which unregisters
LUN devices and frees memory.

fsg_common_init() takes pointer to fsg_common structure as an
argument.  If it is NULL function allocates storage otherwise
uses pointed to memory (handy if fsg_common is a field of another
structure or a static variable).

fsg_common_release() will free storage only if
free_storage_on_release is set -- it is initialised by
fsg_common_init(): set if allocation was done, unset
otherwise (one may overwrite it of course).
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Using version of fsg_buffhd structure with buf field being an
array of characters with predefined size.  Since mass storage
function does not define changing buffer size on run-time it is
not required for the field to be a pointer to void and allocating
space dynamically.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

In the final version, many fsg_dev structures will (be able to)
refer to a single fsg_common structure and so it is required
to move common data to another object which can be shared.

Situation where many fsg_dev structures refer single fsg_common
structure is when a single instance of MSF is used in several
USB configurations.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Removed code that was included when CONFIG_USB_FILE_STORAGE_TEST
was defined.  If this functionality is required one may still use
the original File-backed Storage Gadget.  It has been agreed that
testing functionality is not required in the composite function.

Also removed fsg_suspend() and fsg_resume() which were no
operations.

Moreover, storage_common.c has been modified in such a way that
defining certain macros skips parts of the file.  Those macros
are:
* FSG_NO_INTR_EP -- skips interrupt endpoint descriptors
* FSG_NO_DEVICE_STRINGS -- skips certain strings
* FSG_NO_OTG -- skips OTG descriptor
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Copied file_storage.c to f_mass_storage.c which will be used as
template for the Mass Storage composite Function.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Since storage_common.c no longer references mod_data object
it is now possible to include it before mod_data object is
defined.  This makes it possible to move some defines that
are used as default values of mod_data fields to be defined
in storage_common.c file (where they should be set from
the beginning).

Also, show_ro(), show_file(), store_ro() and store_file()
have been moved to storage_common.c with LUN's device's
drvdata changed from pointing to fsg_dev to pointing to
rw_semaphore (&fsg->filesem).
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

removable and cdrom flag has been added to the fsg_lun structure
removing any references to mod_data object from storage_common.c.

As of read-only flag, previously it was set if a read-only
backing file was specified (which is good) and remained set
even after the file has been closed (which may be considered an
issue).  Currently, the initial read-only flag is preserved so
if it was unset each time file is opened code will try to open
it read-write even if previous file was opened read-only.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Prefixed some identifiers that were defined in storage_common.c file
with "fsg_".  Not all identifiers were prefixed but the ones that are
most likely to produce conflicts when used with other USB functions.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Moved parts of the file_storage.c file which do not reference fsg_dev
structure to newly created storage_common.c file. dump_msg() and
dump_cdb() have been changed to macros to remove fsg_dev reference.
Signed-off-by: NMichal Nazarewicz <m.nazarewicz@samsung.com>
Cc: David Brownell <dbrownell@users.sourceforge.net>
Cc: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

While Linux provided an O_SYNC flag basically since day 1, it took until
Linux 2.4.0-test12pre2 to actually get it implemented for filesystems,
since that day we had generic_osync_around with only minor changes and the
great "For now, when the user asks for O_SYNC, we'll actually give
O_DSYNC" comment.  This patch intends to actually give us real O_SYNC
semantics in addition to the O_DSYNC semantics.  After Jan's O_SYNC
patches which are required before this patch it's actually surprisingly
simple, we just need to figure out when to set the datasync flag to
vfs_fsync_range and when not.

This patch renames the existing O_SYNC flag to O_DSYNC while keeping it's
numerical value to keep binary compatibility, and adds a new real O_SYNC
flag.  To guarantee backwards compatiblity it is defined as expanding to
both the O_DSYNC and the new additional binary flag (__O_SYNC) to make
sure we are backwards-compatible when compiled against the new headers.

This also means that all places that don't care about the differences can
just check O_DSYNC and get the right behaviour for O_SYNC, too - only
places that actuall care need to check __O_SYNC in addition.  Drivers and
network filesystems have been updated in a fail safe way to always do the
full sync magic if O_DSYNC is set.  The few places setting O_SYNC for
lower layers are kept that way for now to stay failsafe.

We enforce that O_DSYNC is set when __O_SYNC is set early in the open path
to make sure we always get these sane options.

Note that parisc really screwed up their headers as they already define a
O_DSYNC that has always been a no-op.  We try to repair it by using it for
the new O_DSYNC and redefinining O_SYNC to send both the traditional
O_SYNC numerical value _and_ the O_DSYNC one.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Grant Grundler <grundler@parisc-linux.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Andreas Dilger <adilger@sun.com>
Acked-by: NTrond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: NKyle McMartin <kyle@mcmartin.ca>
Acked-by: NUlrich Drepper <drepper@redhat.com>
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NJan Kara <jack@suse.cz>

This patch (as1233) makes g_file_storage use the "unaligned" accessors.
This is based on work originally done by Harvey Harrison.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Acked-by: NHarvey Harrison <harvey.harrison@gmail.com>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch (as1231) fixes a use-after-free bug in g_file_storage.  A
device's name may not be available after the device is unregistered,
even if the device structure itself is still allocated.  Since
close_backing_file() prints a LUN's name for debugging, it shouldn't
be called after the LUN has been unregistered.

That whole area needed to be cleaned up; the backing files were
getting closed in a couple of different places.  The patch fixes
things so that they get closed in just one place, as part of the
unbind procedure, immediately before the LUN is unregistered.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
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>

Traditionally, changes to struct file->f_flags have been done under BKL
protection, or with no protection at all.  This patch causes all f_flags
changes after file open/creation time to be done under protection of
f_lock.  This allows the removal of some BKL usage and fixes a number of
longstanding (if microscopic) races.
Reviewed-by: NChristoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: NJonathan Corbet <corbet@lwn.net>

This patch (as1220) automatically disables stalls when g_file_storage
finds itself running with an Atmel device controller, because the
Atmel hardware/driver isn't capable of halting bulk endpoints
correctly.
Reported-by: NStanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Acked-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

In some obscure scenarios e.g. passing a 0-byte backing file
storage, wait_for_completion() would wait forever in fsg_cleanup().

Prevent it by completing the thread in fsg_bind() error path.
Signed-off-by: NFelipe Balbi <felipe.balbi@nokia.com>
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch (as1172) adds the ability to emulate a CD-ROM drive to
g_file_storage.  The emulation is limited, since it presents as a disc
containing a single data track and no audio tracks.  Still, it may
come in useful on occasion.
Signed-off-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Fsync currently has a fdatawrite/fdatawait pair around the method call,
and a mutex_lock/unlock of the inode mutex.  All callers of fsync have
to duplicate this, but we have a few and most of them don't quite get
it right.  This patch adds a new vfs_fsync that takes care of this.
It's a little more complicated as usual as ->fsync might get a NULL file
pointer and just a dentry from nfsd, but otherwise gets afile and we
want to take the mapping and file operations from it when it is there.

Notes on the fsync callers:

 - ecryptfs wasn't calling filemap_fdatawrite / filemap_fdatawait on the
   	lower file
 - coda wasn't calling filemap_fdatawrite / filemap_fdatawait on the host
	file, and returning 0 when ->fsync was missing
 - shm wasn't calling either filemap_fdatawrite / filemap_fdatawait nor
   taking i_mutex.  Now given that shared memory doesn't have disk
   backing not doing anything in fsync seems fine and I left it out of
   the vfs_fsync conversion for now, but in that case we might just
   not pass it through to the lower file at all but just call the no-op
   simple_sync_file directly.

[and now actually export vfs_fsync]
Signed-off-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>

Offer a "how much VBUS power to request" configuration option
for USB gadgets that aren't using board-specific customization
of their gadget or (composite) configuration drivers.

Also remove a couple pointless "depends on USB_GADGET" bits
from the Kconfig text; booleans inside an "if USB_GADGET" will
already have that dependency.

Based on a patch from Justin Clacherty.
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Cc: Justin Clacherty <justin@redfish-group.com>
Tested-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Try to workaround issues with bad SCSI implementations
by ignoring the command size error.
Signed-off-by: NFelipe Balbi <felipe.balbi@nokia.com>
Acked-by: NAlan Stern <stern@rowland.harvard.edu>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Change how the file storage gadget driver builds:  don't
use separate compilation, since it works poorly when key
parts are library code (with init sections etc).  Instead
be as close as we can to "gcc --combine ...".
Signed-off-by: NDavid Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

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>

The bus_id field is going away, use the dev_set_name() function
to set it properly.
Signed-off-by: NKay Sievers <kay.sievers@vrfy.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The bus_id field is going away, use the dev_name() function instead.
Signed-off-by: NKay Sievers <kay.sievers@vrfy.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Gadget tells controller driver to ignore Clear-Feature(HALT_ENDPOINT)
Signed-off-by: NDavid Lopo <lopo.david@gmail.com>
Acked-by: NAlan Stern <stern@rowland.harvard.edu>