staging: ccree: make stub function static inline

The debugfs interface defines stub function if debugfs is not
enabled, which were missing the 'static inline' qualifiers causing
sparse warnings.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: add missing include

Add the missing include of include file with function declarations.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: remove unneeded includes

Remove include files not needed for compilation.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: use a consistent file naming convention

The ccree driver source files were using an inconsistent
naming convention stemming from what the company was called
when they were added.

Move to a single consistent naming convention for better
code readability.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: remove LIBCFS_ALLOC, LIBCFS_FREE and related macros.

LIBCFS_ALLOC
LIBCFS_ALLOC_ATOMIC
LIBCFS_ALLOC_POST
LIBCFS_CPT_ALLOC
LIBCFS_FREE

are no longer used, and so are removed.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: replace LIBCFS_CPT_ALLOC()

LIBCFS_APT_ALLOC() calls kvmalloc_node() with GFP_NOFS
which is not permitted.
Mostly, a kmalloc_node(GFP_NOFS) is appropriate, though occasionally
the allocation is large and GFP_KERNEL is acceptable, so
kvmalloc_node() can be used.

This patch introduces 4 alternatives to LIBCFS_CPT_ALLOC():
 kmalloc_cpt()
 kzalloc_cpt()
 kvmalloc_cpt()
 kvzalloc_cpt().

Each takes a size, gfp flags, and cpt number.

Almost every call to LIBCFS_CPT_ALLOC() passes lnet_cpt_table()
as the table.  This patch embeds that choice in the k*alloc_cpt()
macros, and opencode kzalloc_node(..., cfs_cpt_spread_node(..))
in the one case that lnet_cpt_table() isn't used.

When LIBCFS_CPT_ALLOC() is replaced, the matching LIBCFS_FREE()
is also replaced, with with kfree() or kvfree() as appropriate.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: opencode LIBCFS_ALLOC_ATOMIC calls.

Just call kzalloc(GFP_ATOMIC) directly.
We don't need the warning on failure.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: cfs_percpt_alloc: use kvmalloc(GFP_KERNEL)

this allocation is called from several places, but all are
during initialization, so GFP_NOFS is not needed.
So use kvmalloc and GFP_KERNEL.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: use kmalloc for allocating ksock_tx

The size of the data structure is primarily controlled
by the iovec size, which is limited to 256.
Entries in this vector are 12 bytes, so the whole
will always fit in a page.
So it is safe to use kmalloc (kvmalloc not needed).
So replace LIBCFS_ALLOC with kmalloc.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: lnet-route: use kmalloc for small allocation

This allocation is reasonably small.
As the function is called "*_locked", it might not be safe
to perform a GFP_KERNEL allocation, so be safe and
use GFP_NOFS.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: more conversions to GFP_KERNEL allocations.

These are not called from filesystem context, so use
GFP_KERNEL, not LIBCFS_ALLOC().

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: more LIBCFS_ALLOC conversions to GFP_KERNEL allocations.

None of these need GFP_NOFS so allocate directly.
Change matching LIBCFS_FREE() to kfree() or kvfree().

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: Convert more LIBCFS_ALLOC allocation to direct GFP_KERNEL

None of these need to be GFP_NOFS, so use GFP_KERNEL explicitly
with kmalloc(), kvmalloc(), or kvmalloc_array().
Change matching LIBCFS_FREE() to kfree() or kvfree()

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: change some LIBCFS_ALLOC calls to k?alloc(GFP_KERNEL)

When an allocation happens from process context rather than
filesystem context, it is best to use GFP_KERNEL rather than
LIBCFS_ALLOC() which always uses GFP_NOFS.
This include initialization during, or prior to, mount,
and code run from separate worker threads.

So for some of these cases, switch to kmalloc, kvmalloc, or
kvmalloc_array() as appropriate.
In some cases we preserve __GFP_ZERO (via kzalloc/kvzalloc), but in
others it is clear that allocated memory is immediately initialized.

In each case, the matching LIBCFS_FREE() is converted to
kfree() or kvfree()

This is just a subset of locations that need changing.
As there are quite a lot, I've broken them up into several
ad-hoc sets to avoid review-fatigue.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: lnet: use kmalloc/kvmalloc in router_proc

The buffers allocated in router_proc are to temporarily
hold strings created for procfs files.
So they do not need to be zeroed and are safe to use
GFP_KERNEL.
So use kmalloc() directly except in two cases where it
isn't trivial to confirm that the size is always small.
In those cases, use kvmalloc().

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: lnet: selftest: don't allocate small strings.

All of the "name" buffers here are at most LST_NAME_SIZE+1
bytes, so 33 bytes at most.
They are only used temporarily during the life of the function
that allocates them.
So it is much simpler to just allocate on the stack.
Worst case is lst_tet_add_ioct(), which allocates
3 for these which 99 bytes on the stack, instead of the 24 that would
have been allocated for 64-bit pointers.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: lnet: switch to cpumask_var_t

So that we can use the common cpumask allocation functions,
switch to cpumask_var_t.
We need to be careful not to free a cpumask_var_t until the
variable has been initialized, and it cannot be initialized
directly.
So we must be sure either that it is filled with zeros, or
that zalloc_cpumask_var() has been called on it.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: replace simple cases of LIBCFS_ALLOC with kzalloc.

All usages of the form
  LIBCFS_ALLOC(variable, sizeof(variable))
or
  LIBCFS_ALLOC(variable, sizeof(variable's-type))

are changed to
  variable = kzalloc(sizeof(...), GFP_NOFS);

Similarly, all
   LIBCFS_FREE(variable, sizeof(variable))
become
   kfree(variable);

None of these need the vmalloc option, or any of the other minor
benefits of LIBCFS_ALLOC().

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: pi433: replace shifting with BIT macro

Fixes checkpatch warnings:

  CHECK: Prefer using the BIT macro

Signed-off-by: Valentin Vidic <Valentin.Vidic@CARNet.hr>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: greybus: camera: cleanup multiple checks for null pointers

Fixed coding style issue regarding null comparison at multiple lines.
Issue reported by checkpatch.pl

Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Acked-by: Johan Hovold <johan@kernel.org>
Reviewed-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: android: ion: Switch from WARN to pr_warn

Syzbot reported a warning with Ion:

WARNING: CPU: 0 PID: 3502 at drivers/staging/android/ion/ion-ioctl.c:73 ion_ioctl+0x2db/0x380 drivers/staging/android/ion/ion-ioctl.c:73
Kernel panic - not syncing: panic_on_warn set ...

This is a warning that validation of the ioctl fields failed. This was
deliberately added as a warning to make it very obvious to developers that
something needed to be fixed. In reality, this is overkill and disturbs
fuzzing. Switch to pr_warn for a message instead.

Reported-by: syzbot+fa2d5f63ee5904a0115a@syzkaller.appspotmail.com
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: android: ion: Add __GFP_NOWARN for system contig heap

syzbot reported a warning from Ion:

  WARNING: CPU: 1 PID: 3485 at mm/page_alloc.c:3926

  ...
   __alloc_pages_nodemask+0x9fb/0xd80 mm/page_alloc.c:4252
  alloc_pages_current+0xb6/0x1e0 mm/mempolicy.c:2036
  alloc_pages include/linux/gfp.h:492 [inline]
  ion_system_contig_heap_allocate+0x40/0x2c0
  drivers/staging/android/ion/ion_system_heap.c:374
  ion_buffer_create drivers/staging/android/ion/ion.c:93 [inline]
  ion_alloc+0x2c1/0x9e0 drivers/staging/android/ion/ion.c:420
  ion_ioctl+0x26d/0x380 drivers/staging/android/ion/ion-ioctl.c:84
  vfs_ioctl fs/ioctl.c:46 [inline]
  do_vfs_ioctl+0x1b1/0x1520 fs/ioctl.c:686
  SYSC_ioctl fs/ioctl.c:701 [inline]
  SyS_ioctl+0x8f/0xc0 fs/ioctl.c:692

This is a warning about attempting to allocate order > MAX_ORDER. This
is coming from a userspace Ion allocation request. Since userspace is
free to request however much memory it wants (and the kernel is free to
deny its allocation), silence the allocation attempt with __GFP_NOWARN
in case it fails.

Reported-by: syzbot+76e7efc4748495855a4d@syzkaller.appspotmail.com
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Laura Abbott <labbott@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: gs_fpgaboot: remove FSF's mailing address from io.h

This patch removes FSF's mailing address issue from io.h found by
checkpatch.pl tool.

Signed-off-by: Shubham Kumaram <shubhamkumaram@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: sm750fb: Remove multiple blank lines

Remove multiple blank lines. Issue found by checkpatch.

Signed-off-by: Marcelo Guzmán Lamperti <marcelo.guzman.lamperti@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: fsl-dpaa2/eth: Use affine DPIO services

Use the newly added DPIO service API to map cpu-affine DPIO services
to channels.

The DPAA2 Ethernet driver already had mappings of frame queues and
channels to cpus, but had no control over the DPIOs used. We can
now ensure full affinity of hotpath hardware resources to cores,
which improves performance and almost eliminates some resource
contentions (e.g. enqueue/dequeue busy counters should be close to
zero from now on).

Making the pull channel operation core affine brings the most
significant benefits. This ensures the same DPIO service will be
used for all dequeue commands issued for a certain frame queue,
which is in line with the way hardware is optimized.

Additionally, we also use affine DPIOs for the frame enqueue and
buffer release operations in order to avoid resource contention.
dpaa2_io_service_register() and dpaa2_io_service_rearm()
functions receive an affine DPIO as argument mostly for uniformity,
but this doesn't change the previous functionality.

Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: fsl-mc/dpio: Add dpaa2_io_service_select() API

All DPIO service API functions receive a dpaa2_io service pointer
as parameter (NULL meaning any service will do) which indicates
the hardware resource to be used to execute the specified command.

There isn't however any available API for obtaining such a service
reference that could be used further, effectively forcing the users
to always request a random service for DPIO operations.
(The DPIO driver holds internally an array mapping services to cpus,
and affine services can be indirectly requested by a couple of API
functions: dpaa2_io_service_register and dpaa2_io_service_rearm
use the cpu id provided by the user to select the corresponding
service)

This patch adds a function for selecting a DPIO service based on
the specified cpu id. If the user provides a "don't care" value
for the cpu, we revert to the default behavior and return the next
DPIO, taken in a round-robin fashion from a list of available
services.

Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Acked-by: Roy Pledge <roy.pledge@nxp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: vt6656: Fix unnecessary 'out of memory' message

This patch fixes one of the warnings as noted by checkpatch.pl related
to unnecessary 'out of memory' message.

This patch fixes the following checkpatch.pl error:

WARNING: Possible unnecessary 'out of memory' message

Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>

staging: rtl8192u: Replace mdelay with msleep in rtl8192_usb_probe

rtl8192_usb_probe is not called in an interrupt handler
nor holding a spinlock.
The function mdelay in it can be replaced with msleep,
to avoid busy wait.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>

staging: comedi: adv_pci1760: fix typo in comments

Changed 'firware' to 'firmware'

Signed-off-by: Richard Sudaryono <richardsudaryono@gmail.com>

Staging: rtlwifi: Remove unused variable and the code

Remove unused variable and also remove unused code
associated with initializing the unused variable.

Unused variable was detected using the following
semantic patch by coccinelle.

@@
type T;
identifier i;
constant C;
@@

(
extern T i;
|
- T i;
  <+... when != i
- i = C;
  ...+>
)

Signed-off-by: Shreeya Patel <shreeya.patel23498@gmail.com>

staging: vc05_services: fix checkpatch.pl errors

Fix some errors for wrong brace position reported by checkpatch.

Signed-off-by: Sidong Yang <realwakka@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vc04_services: Prefer WARN_ON_ONCE instead of if condition followed by BUG.

Coccinelle suggested to use BUG_ON instead of if condition followed by BUG
but BUG_ON should be used in situations where integrity of the system is no
longer guaranteed. In this case, as suggested by Stefan Wahren, vchiq isn't
critical.
Since it is not critical, BUG_ON should be avoided.
Replaced if condition followed by BUG with WARN_ON_ONCE.

Signed-off-by: Kishore KP <kishore.p@techveda.org>
Signed-off-by: Suniel Mahesh <sunil.m@techveda.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vc04_services: Fix platform_no_drv_owner.cocci warnings.

Removed .owner field initialization, platform core does it automatically.
Pointed out by Coccinelle.

Signed-off-by: Kishore KP <kishore.p@techveda.org>
Signed-off-by: Suniel Mahesh <sunil.m@techveda.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: wlan-ng: hfa384x_usb: fixed two line limit coding style issues

Fixed two coding style issues.

Signed-off-by: Andy Pusch <drag@black-pixel.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: vme: vme_user: fixed an alignment coding style issue

Fixed a coding style issue.

Signed-off-by: Philippe Loctaux <loctauxphilippe@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: android: check the return value of register_shrinker

register_shrinker call is made in ashmem_init, it may return error code,
so we need to check it.

Signed-off-by: Xiongwei Song <sxwjean@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: android: check for error from register_shrinker in ion_heap_init_shrinker

The function register_shrinker in ion_heap_init_shrinker may return an
error, check it out. Meanwhile, ion_heap_init_shrinker has to a return
value.

Signed-off-by: Xiongwei Song <sxwjean@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: android: ashmem: add procfs fdinfo helper

This print the inode number of backing file and the name in
/proc/pid/fdinfo/fd.
These information helps users to know which processes are sharing the same
ashmem.

Signed-off-by: Zhai Zhaoxuan <kxuanobj@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: most: convert macro to static function

This fixes checkpatch warning:
CHECK: Macro argument reuse 'buf' - possible side effects?

Signed-off-by: George Edward Bulmer <gebulmer@googlemail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: most: Fix identifiers to function parameters

fixed "function definition argument should have an identifier name",
with appropriate identifier names. Pointed out by checkpatch.

Signed-off-by: Ravi Eluri <venkataravi.e@techveda.org>
Signed-off-by: Suniel Mahesh <sunil.m@techveda.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: most: Avoid trailing semicolon for macros

Fixes checkpatch warning:
macros should not use a trailing semicolon.

Signed-off-by: Ravi Eluri <venkataravi.e@techveda.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: pi433 - fix ccheckpatch issue, updated include line.

Updated the include of compat.h to fix checkpatch error

Signed-off-by: Derek Robson <robsonde@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: pi433: fix CamelCase for payload identifiers

Fixes checkpatch warnings:

  CHECK: Avoid CamelCase: <payloadLength>
  CHECK: Avoid CamelCase: <payloadReady>

Signed-off-by: Valentin Vidic <Valentin.Vidic@CARNet.hr>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: pi433: fix CamelCase for maxMinus variables

Fixes checkpatch warnings:

  CHECK: Avoid CamelCase: <maxMinus12>
  CHECK: Avoid CamelCase: <maxMinus24>
  CHECK: Avoid CamelCase: <maxMinus36>
  CHECK: Avoid CamelCase: <maxMinus48>
  CHECK: Avoid CamelCase: <maxMinus6>

Signed-off-by: Valentin Vidic <Valentin.Vidic@CARNet.hr>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: pi433: remove unused rf69_reset_flag function

Function rf69_reset_flag is unused and should be removed.

Signed-off-by: Marcin Ciupak <marcin.s.ciupak@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: pi433: remove unused rf69_set_sync_tolerance function

Function rf69_set_sync_tolerance is unused and should be removed.

Signed-off-by: Marcin Ciupak <marcin.s.ciupak@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: update TODO

Update TODO to reflect work done

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

stating: ccree: fix allocation of void sized buf

We were allocating buffers using sizeof(*struct->field) where field was
type void.  Fix it by having a local variable with the real type.

Cc: stable@vger.kernel.org
Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: put pointer next to var name

Put pointer next to var name as per coding style.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: fold common code into service func

Fold common code in hash call into service functions.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: fix indentation of func params

Fix indentation of some function params in hash code for
better readability.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: do not map bufs in ahash_init

hash_init was mapping DMA memory that were then being unmap in
hash_digest/final/finup callbacks, which is against the Crypto API
usage rules (see discussion at
https://www.mail-archive.com/linux-crypto@vger.kernel.org/msg30077.html)

Fix it by moving all buffer mapping/unmapping or each Crypto API op.

This also properly deals with hash_import() not knowing if
hash_init was called or not as it now no longer matters.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: allocate hash bufs inside req ctx

Move to allocating the buffers needed for requests as part of
the request structure instead of malloc'ing each one on it's
own, making for simpler (and more efficient) code.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: use array for double buffer

The ccree hash code is using a double buffer to hold data
for processing but manages the buffers and their associated
data count in two separate fields and uses a predicate to
chose which to use.

Move to using a proper 2 members array for a much cleaner code.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: remove unused field

Remove unused struct field.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: use Makefile to include PM code

Replace ugly ifdefs with some inline macros and Makefile magic
for optionally including power management related code for
better readability.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: handle end of sg list gracefully

If we are asked for number of entries of an offset bigger than the
sg list we should not crash.

Cc: stable@vger.kernel.org
Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: check DMA pool buf !NULL before free

If we ran out of DMA pool buffers, we get into the unmap
code path with a NULL before. Deal with this by checking
the virtual mapping is not NULL.

Cc: stable@vger.kernel.org
Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: failing the suspend is not an error

PM suspend returning a none zero value is not an error. It simply
indicates a suspend is not advised right now so don't treat it as
an error.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

stating: ccree: revert "staging: ccree: fix leak of import() after init()"

This reverts commit c5f39d07860c ("staging: ccree: fix leak of import()
after init()") and commit aece09024414 ("staging: ccree: Uninitialized
return in ssi_ahash_import()").

This is the wrong solution and ends up relying on uninitialized memory,
although it was not obvious to me at the time.

Cc: stable@vger.kernel.org
Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: add backlog processing

Crypto API tfm providers are required to provide a backlog
service, if so indicated, that queues up requests in the case
of the provider being busy and processing them later.

The ccree driver did not provide this facility. Add it now.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: break send_request and fix ret val

The send_request() function was handling both synchronous
and asynchronous invocations, but were not handling
the asynchronous case, which may be called in an atomic
context, properly as it was sleeping.

Start to fix the problem by breaking up the two use
cases to separate functions calling a common internal
service function and return error instead of sleeping
for the asynchronous case.

The next patch will complete the fix by implementing
proper backlog handling.

Fixes: abefd6741d ("staging: ccree: introduce CryptoCell HW driver").
Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: remove unused leftover field

Remove the unused monitor_desc field.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: tag debugfs init/exit func properly

The debugfs global init and exit functions were missing
__init and __exit tags, potentially wasting memory.
Fix it by properly tagging them.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: copy larval digest from RAM

The ccree driver was using a DMA operation to copy larval digest
from the ccree SRAM to RAM. Replace it with a simple memcpy.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: pick alloc mem flags based on req flags

The ccree driver was allocating memory using GFP_KERNEL flag
always, ignoring the flags set in the crypto request. Fix it
by choosing gfp flags based on crypto request flags.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: remove GFP_DMA flag from mem allocs

Remove bogus GFP_DMA flag from memory allocations. ccree driver
does not operate over an ISA or similar limited bus.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: fold reg common defines into driver

Fold the 2 macro defined in dx_reg_common.h into the file they
are used in and delete the file.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: fold hash defs into queue defs

Fold the two remaining enum in hash defs into the queue defs
that are using them and delete the hash defs include file.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: SPDXify driver

Replace verbatim GPL v2 copy with SPDX tag.

Signed-off-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: ccree: mark debug_regs[] as static

The global array clashes with an existing symbol of the same name:

drivers/staging/ccree/cc_debugfs.o:(.data+0x0): multiple definition of `debug_regs'
drivers/mmc/host/s3cmci.o:(.data+0x70): first defined here

We should fix both, this one addresses the ccree driver by removing
the symbol from the global namespace.

Fixes: 9bdd203b4dc8 ("s3cmci: add debugfs support for examining driver and hardware state")
Fixes: b3ec9a6736f2 ("staging: ccree: staging: ccree: replace sysfs by debugfs interface")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Gilad Ben-Yossef <gilad@benyossef.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vboxvideo: Fix incorrect type in assignment sparse warning

Sparse emitted the following warning:
../drivers/staging/vboxvideo/vbox_fb.c:173:27: warning: incorrect type in assignment (different address spaces)
../drivers/staging/vboxvideo/vbox_fb.c:173:27:    expected char [noderef] <asn:2>*screen_base
../drivers/staging/vboxvideo/vbox_fb.c:173:27:    got void *virtual

The vbox_bo buffer object kernel mapping is handled by a call
to ttm_bo_kmap() prior to the assignment of bo->kmap.virtual to
info->screen_base of type char __iomem*.
Casting bo->kmap.virtual to char __iomem* in this assignment fixes
the warning.

vboxvideo: Fix address space of expression removal sparse warning

Sparse emitted the following warning:
../drivers/staging/vboxvideo/vbox_main.c:64:25: warning: cast removes address space of expression

vbox->vbva_buffers iomapping is handled by calling vbox_accel_init()
from vbox_hw_init().
__force attribute is used in assignment to vbva to fix the warning.

Signed-off-by: Alexander Kapshuk <alexander.kapshuk@gmail.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/fbtft: Remove unnecessary braces from if/else

The Linux kernel coding style states that braces should only be used
when necessary.

This fixes the checkpatch warning

WARNING: line over 80 characters
+ } else if (display->regwidth == 8 && display->buswidth == 9 && par->spi) {

introduced by patch #1.

Signed-off-by: Luis Gerhorst <linux-kernel@luisgerhorst.de>
Acked-by: Jonny Schaefer <schaefer.jonny@gmail.com>
Acked-by: Alexander Wuerstlein <arw@cs.fau.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/fbtft: Remove newline after else in else-if

This removes the following warning issued by checkpatch

WARNING: suspect code indent for conditional statements (8, 8)
+ } else
+ if (display->regwidth == 8 && display->buswidth == 9 && par->spi) {

Signed-off-by: Luis Gerhorst <linux-kernel@luisgerhorst.de>
Acked-by: Jonny Schaefer <schaefer.jonny@gmail.com>
Acked-by: Alexander Wuerstlein <arw@cs.fau.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: lustre: Fix prefer kcalloc over kzalloc with multiply

Use kcalloc for allocating an array instead of kzalloc with
multiply. kcalloc is the preferred API. Issue reported by
checkpatch.pl

Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: lustre: Fix prefer seq_puts to seq_printf

Use seq_puts() for strings without format specifiers instead of
seq_printf(). Issue reported by checkpatch.pl

Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers: lustre: obdclass: simplify unregister_shrinker() usage

lu_global_fini() explicitly uses knowledge about shrinker's
internals to make decision about calling of unregister_shrinker().
Now this check was integrated into unregister_shrinker(),
so it is safe to call it against unregistered shrinker.

Signed-off-by: Aliaksei Karaliou <akaraliou.dev@gmail.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: lnet: discard CFS_ALLOC_PTR

These trivial wrappers hurt readability and
as they use kvmalloc, they are overly generic.

So discard them and use kmalloc/kfree as is
normal in Linux.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: lnet-lib: opencode some alloc/free functions.

These functions just call LIBCFS_ALLOC() which in-turn
calls kvmalloc().
In none of these cases is the 'vmalloc' option needed.

LIBCFS_ALLOC also produces a warning if NULL is returned,
but that can be provided with CONFIG_SLAB_DEBUG.

LIBCFS_ALLOC zeros the memory, so we need to use
__GFP_ZERO too.

So with one exception where the alloc function is not trivial,
open-code the alloc and free functions using kmalloc and kfree.

Note that the 'size' used in lnet_md_alloc() is limited and less than
a page because LNET_MAX_IOV is 256, so kvmalloc is not needed.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: libcfs: remove prng

The cfs prng is no longer used, so discard it.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: replace cfs_get_random_bytes calls with get_random_byte()

The cfs_get_random_bytes() interface adds nothing of value
to get_random_byte() (which it uses internally).  So just use the
standard interface.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: replace cfs_srand() calls with add_device_randomness().

The only places that cfs_srand is called, the random bits are
mixed with bits from get_random_bytes().  So it is equally effective
to add entropy to either pool.
So we can replace calls to cfs_srand() with calls that add the
entropy with add_device_randomness().  That function adds time-based
entropy, so we can discard the ktime_get_ts64 calls.

One location in lustre_handles.c only adds timebased
entropy.  This cannot improve the entropy provided by get_random_bytes(),
so just discard that call.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: replace cfs_rand() with prandom_u32_max()

All occurrences of
   cfs_rand() % X
are replaced with
   prandom_u32_max(X)

cfs_rand() is a simple Linear Congruential PRNG.  prandom_u32_max()
is at least as random, is seeded with more randomness, and uses
cpu-local state to avoid cross-cpu issues.

This is the first step is discarding the libcfs prng with
the standard linux prng.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: libcfs: remove wi_data from cfs_workitem

In every case, the value passed via wi_data can be determined
from the cfs_workitem pointer using container_of().

So use container_of(), and discard wi_data.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: libcfs: use a workqueue for rehash work.

lustre has a work-item queuing scheme that provides the
same functionality as linux work_queues.
To make the code easier for linux devs to follow, change
to use work_queues.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: obd_mount: fix possible race with module unload.

lustre_fill_super() calls client_fill_super() without holding a
reference to the module containing client_fill_super.  If that
module is unloaded at a bad time, this can crash.

To be able to get a reference to the module using
try_get_module(), we need a pointer to the module.

So replace
  lustre_register_client_fill_super() and
  lustre_register_kill_super_cb()
with a single
  lustre_register_super_ops()
which also passed a module pointer.

Then use a spinlock to ensure the module pointer isn't removed
while try_module_get() is running, and use try_module_get() to
ensure we have a reference before calling client_fill_super().

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: obdclass: remove vfsmount option from client_fill_super

This arg is always NULL and is never used.
So discard it from this and related functions.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: obdclass: remove pointless struct lustre_mount_data2

This is used to pass a void* and NULL to lustre_fill_super().
It is easier just to pass the void*.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: obdclass: discard FS_NEEDS_DEV flag.

Lustre mounts do not need a dev, as we can see from lustre_mount()
calling mount_nodev().  So remove the flag which could cause
confusion elsewhere.

Also format lustre_fs_types so that fields line up.

Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: disable preempt while sampling processor id.

Calling smp_processor_id() without disabling preemption
triggers a warning (if CONFIG_DEBUG_PREEMPT).
I think the result of cfs_cpt_current() is only used as a hint for
load balancing, rather than as a precise and stable indicator of
the current CPU.  So it doesn't need to be called with
preemption disabled.

So disable preemption inside cfs_cpt_current() to silence the warning.

Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: lustre: use strim instead of cfs_trimwhite.

Linux lib provides identical functionality to cfs_trimwhite,
so discard that code and use the standard.

Signed-off-by: NeilBrown <neilb@suse.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iio: chemical: ccs811: Fix output of IIO_CONCENTRATION channels

in_concentration_raw should report, according to sysfs-bus-iio documentation,
a "Raw (unscaled no offset etc.) percentage reading of a substance."

Modify scale to convert from ppm/ppb to percentage:
1 ppm = 0.0001%
1 ppb = 0.0000001%

There is no offset needed to convert the ppm/ppb to percentage,
so remove offset from IIO_CONCENTRATION (IIO_MOD_CO2) channel.

Cc'd stable to reduce chance of userspace breakage in the long
run as we fix this wrong bit of ABI usage.

Signed-off-by: Narcisa Ana Maria Vasile <narcisaanamaria12@gmail.com>
Cc: <Stable@vger.kernel.org>
Reviewed-by: Matt Ranostay <matt.ranostay@konsulko.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

imu:adis16480: fix debugfs_simple_attr.cocci warnings

Use DEFINE_DEBUGFS_ATTRIBUTE rather than DEFINE_SIMPLE_ATTRIBUTE
for debugfs files.

Semantic patch information:
Rationale: DEFINE_SIMPLE_ATTRIBUTE + debugfs_create_file()
imposes some significant overhead as compared to
DEFINE_DEBUGFS_ATTRIBUTE + debugfs_create_file_unsafe().

Generated by: scripts/coccinelle/api/debugfs/debugfs_simple_attr.cocci

Signed-off-by: Venkat Prashanth B U <venkat.prashanth2498@gmail.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

gyro:adis16136: fix debugfs_simple_attr.cocci warnings

Use DEFINE_DEBUGFS_ATTRIBUTE rather than DEFINE_SIMPLE_ATTRIBUTE
for debugfs files.

Semantic patch information:
Rationale: DEFINE_SIMPLE_ATTRIBUTE + debugfs_create_file()
imposes some significant overhead as compared to
DEFINE_DEBUGFS_ATTRIBUTE + debugfs_create_file_unsafe().

Generated by: scripts/coccinelle/api/debugfs/debugfs_simple_attr.cocci

Signed-off-by: Venkat Prashanth B U <venkat.prashanth2498@gmail.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: iio: light: Add breaks to lengthy lines

This fixes three instances of checkpatch warning:
WARNING: line over 80 characters

Signed-off-by: George Edward Bulmer <gebulmer@googlemail.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iio: imu: st_lsm6dsx: remove st_lsm6dsx_write_with_mask() declaration

Remove st_lsm6dsx_write_with_mask() declaration since it has been removed
in commit 6674bef628e6 ("iio: imu: st_lsm6dsx: add regmap API support")

Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iio: imu: st_lsm6dsx: pre-allocate read buffer at bootstrap

Allocate device read buffer at bootstrap and do not put it on the stack
since it is pretty big (~200B) and its size will increase adding support
to device hw timestamp.
Moreover this patch fixes following sparse warnings:
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c:250:17: warning: Variable length
array is used.
drivers/iio/imu/st_lsm6dsx/st_lsm6dsx_buffer.c:283:55: error: cannot size
expression

Fixes: 290a6ce11d93 ("iio: imu: add support to lsm6dsx driver")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iio: add field identifier for @use_count kernel-doc

Kernel-doc for @use_count does not currently have a field identifier.
All the rest of the fields do. @use_count is used internally and should
not be accessed directly by the driver so it should be marked as so.

Add [INTERN] identifier to @use_count field.

Signed-off-by: Tobin C. Harding <me@tobin.cc>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iio: add kernel-doc for field @owner

When building kernel documentation sphinx emits the following warning

warning: No description found for parameter 'owner'

Add description for struct member 'owner'.

Signed-off-by: Tobin C. Harding <me@tobin.cc>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Staging: iio: cdc: ad7152: Prefer using BIT macro

This patch fixes the following checkpatch.pl issue at multiple lines:

CHECK: Prefer using the BIT macro

Signed-off-by: Sumit Pundir <pundirsumit11@gmail.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>