Trivial fix to spelling mistake in TODO file.
Signed-off-by: Sabin Mihai Rapan <sabin.rapan@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Trivial fix to spelling mistake in README.txt
Signed-off-by: Sabin Mihai Rapan <sabin.rapan@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

In the remaining files that include libcfs_all.h,
replace it with other include files as necessary,
then remove libcfs_all.h
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Again, most of these are not needed.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Most of these aren't needed, a few can be simplified.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

None of these files seem to nee libcfs_all.h
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Some files didn't need it at all, others just needed
one or two includes.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Now no *.h files include libcfs_all.h - only *.c files.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Instead of the catch-all libcfs_all.h, just include the
files actually needed in different places.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Some of these include files didn't need to
include libcfs_all.h.  A couple did so replace with just the
base include files needed.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Other smaller include files are sufficient.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

We want to be more focused in what is included where.
So we remove libcfs.h from where it isn't needed.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Lots of places include libcfs.h, and it includes lots of other include
files.  Many of these aren't needed in many places.  It is tidier and
better documentation to just include what is needed.

So remove all the includes from libcfs.h and create libcfs_all.h which
contains them.  Then change every reference to libcfs.h to instead
include libcfs_all.h

Next several patches will remove that from various files
in small batches
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This file no longer contains enough content
to justify a separate file.  So merge with
libcfs.h.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Use memalloc_noreclaim_save() and memalloc_noreclaim_restore(),
and for testing, just directly test the flag in current->flags
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

These are network related on only used in lnet, so move
to lib-lnet.h
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Both files that use this variable include tracefile.h, and it
seems a more suitable home for the declaration.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This type is only used in libcfs/module.c, so make it local to there.
If any other module ever wanted to add its own symlinks,
it would probably be easiest to export lnet_debugfs_root
and just call debugfs_create_symlink as required.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

There is some confusion with names here - make it all uniform.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

libcfs allows other modules to register handlers for ioctls.
The implementation it uses for this is nearly identical to a
blocking notifier chain, so change to use that.

The biggest difference is that the return value from notifier has a
defined format, where libcfs_register_ioctl uses -EINVAL to mean
"continue".  This requires a little bit of conversion.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cfs_block_sigsinv() and cfs_restore_sigs() are simple
wrappers which save a couple of line of code and
hurt readability for people not familiar with them.
They aren't used often enough to be worthwhile,
so discard them and open-code the functionality.

The sigorsets() call isn't needed as or-ing with current->blocked is
exactly what sigprocmask(SIG_BLOCK) does.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This macro is only used for debug messages, so use
it to the debug code.
Also improve the documentation slightly.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This undocumented macro seems to represent "a small amount of time".
Sometimes it is used as-is, some times it is multiplied by 5 for no
obvious reason.
It does not appear that there is any connection between the different
places it is used - they all just want a short period for different
purposes and of different durarions.

So discard CFS_TICK and lets each use-site just use whatever number
of jiffies seems appropriate in that case.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

This macro is only used once to test if a value is
a power of two.  So use is_power_of_2() instead and
discard the macro.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

lustre only sends 32bits of capabilities in on-the-wire RPC calls.
It current strips off higher bits and uses a 32bit cfs_cap_t
throughout.
Though there is a small memory cost, it is cleaner to use
kernel_cap_t throughout and only truncate when marshalling
data for RPC calls.

So this patch replaces cfs_cap_t with kernel_cap_t throughout,
and where a cfs_cap_t was previous stored in a __u32, we now
store cap.cap[0] instead.

With this, we can remove include/linux/libcfs/curproc.h
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Lustre has a 'squash credentials' concept similar to the "anon_uid"
for nfsd.  When accessing a file with squashed credentials, we
need to also drop capabilities.
Linux has cap_drop_fs_set() and cap_drop_nfsd_set().  Rather than
taking a completely different approach, this patch changes lustre
to use this same cap_drop_*_set() approach.

With this change we also drop CAP_MKNOD and CAP_MAC_OVERRIDE
which are probably appropriate, and don't drop
CAP_SYS_ADMIN or CAP_SYS_BOOT which should be irrelevant for
file permission checking

Calling both cap_drop_*_set() seems a bit clumsy, but gets
the job done.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Just use current->pid and current->comm directly, instead
of having wrappers.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

libcfs_cpu.c manages CPU partitions.  In the !CONFIG_SMP case, most
of this disappears and 'static inline's from libcfs_cpu.h are used.
However we still allocate a 'struct cfs_cpt_table' and keep some
dummy data in it.  This is a bit pointless.

This patch removes all the !CONFIG_SMP code from libcfs_cpu.c and
conditionally compiles the whole file only when CONFIG_SMP.
We no longer allocate a 'struct cfs_cpt_table' on !CONFIG_SMP,
and don't even declare a structure.  The name "cfs_cpt_tab"
becomes always "NULL", which allows some code to be optimized away.

This means that cfs_cpt_tab can sometimes be NULL, so we need to
discard the assertion that it isn't.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

The variable "cfs_cpt_table" has the same name as
the structure "struct cfs_cpt_table".
This makes it hard to use #define to make one disappear
on a uni-processor build, but keep the other.
So rename the variable to cfs_cpt_tab.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

There is no longer any value in having this separate
subdirectory, so promote the files in it.
Also tidy the Makefile a little to use the common "*-obj-y"
macro name.  This will allow individual files to be conditionally
compiled.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Many lustre modules depend on libcfs having initialized
properly, but do not explicit check that it did.
When lustre is built as discrete modules, this does not
cause a problem because if the libcfs module fails
initialization, the other modules don't even get loaded.

When lustre is compiled into the kernel, all module_init()
routines get run, so they need to check the required initialization
succeeded.

This patch splits out the initialization of libcfs into a new
libcfs_setup(), and has all modules call that.

The misc_register() call is kept separate as it does not allocate any
resources and if it fails, it fails hard - no point in retrying.
Other set-up allocates resources and so is best delayed until they
are needed, and can be worth retrying.

Ideally, the initialization would happen at mount time (or similar)
rather than at load time.  Doing this requires each module to
check dependencies when they are activated rather than when
they are loaded.  Achieving that is a much larger job that would
have to progress in stages.

For now, this change ensures that if some initialization in libcfs
fails, other modules will fail-safe.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

A module_init() function that registers the services
of the module should do that last, after all other
initialization has succeeded.
This patch moves the class_register_type() call to the
end and ensures everything else that might have been
set up, is cleaned up on error.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Replace // SPDX-License-Identifier: GPL-2.0+ by
/* SPDX-License-Identifier: GPL-2.0+ */ as per licensing rule
for C header files. Issue found by checkpatch. Part of
Eudyptula Challenge.
Signed-off-by: Nishad Kamdar <nishadkamdar@gmail.com>
Reviewed-by: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Merge tag 'iio-for-4.18b' of git://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio into staging-next

Jonathan writes:

Second set of new device support, features and cleanup for IIO in the 4.18 cycle

Usual mixture of new devices support and other stuff.  A couple of
staging graduations in here and some old driver drops.

New device support
* ad5686
  - Add support for AD5691R, AD5692R, AD5693 and AD5693R i2c DACs
  - Add support for AD5681R, AD5682R, AD5683 and AD5683R SPI DACs
* lmp91000
  - Add ID for LMP91002
* stm32-dfsdm
  - Add support for the stm32mp1 devices.

Drivers dropped
* ADE7753
  - No longer easily available, no users came forward and needs a lot
    of work to move out of staging.
* ADE7754
* ADE7758
* ADE7759

Staging graduations
- ad2s1200 - good cleanup from David Veenstra.
- tsl2772 (was tsl2x7x) - Brian's quest is at an end and
  staging/iio/light is no more!

Features
* tools
  - loop forever on a negative number of loops being specified.
* ltc 2632
  - add of_match_table
  - support an external reference regulator.
* mag3110
  - Support continuous mode when running fast as it increases the
    posssible sampling rate.
* ti-ads8688
  - Add trigger and buffer support to this ADC.

Cleanups / minor tweaks.
* tools
  - tidy up types in iio_generic_buffer.
* ad2s1200
  - Whitespace cleanup.
  - Drop pointless initializations.
  - Improve kernel docs.
  - Reorder to setup the SPI prior to device registration (race fix).
  - Change to modern gpiod framework.
  - Drop platform data and move to DT bindings. There are no in
    kernel users of the platform data. Any out of tree boards will
    need changes.
  - Add an explicit compatible table.
  - Provide _scale for angular velocity and angle channels.
  - Add David Veenstra to copyright notice as this cleanup was non
    trivial.
* ade8366
  - Avoid a race by ensuring channel init is before registration
    with the subsystem.
* afe
  - binding spelling fix.
* imx7d-adc
  - typo fix in Freescale
* inv_mpu6050
  - tidy up an ugly loop to take advantage of known entry condition.
  - add a comment explaining why the sensible sampling frequencies
    are more limited than might be immediately apparent (aliasing).
* mma8452
  - reduce the sleep time on data not ready when running at high
    frequency.
* stm32-dfsdm
  - add missing header.
* tsl2583
  - fix integration_time_availabe which was in microsecs rather the
    millisecs.
* tsl2x7x/tsl2772
  - Whitespace cleanup.
  - Change to direct returns where sensible.
  - Turn the chip off in the registration error path.
  - Use macro to reduce repition when setting up the device_info
    structures.
  - Change the _available attributes over to read_avail callback,
    and the range definitions that supports.
  - Fix some wrong period values.
  - Add some range checkign for _write_raw.
  - Rename the driver to tsl2772 to avoid wild card clash problems
    in future.

Although the driver allows frequencies between 4 and 1000 Hz, only the
frequencies advertised in the available frequencies file are backed
properly by a low-pass filter to prevent aliasing, so it's best to use
them. Since this is not obvious to the user, add a comment explaining
what's going on.
Signed-off-by: Martin Kelly <mkelly@xevo.com>
Acked-by: Jean-Baptiste Maneyrol <jmaneyrol@invensense.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

Signed-off-by: Silvan Murer <silvan.murer@gmail.com>
Reviewed-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

Signed-off-by: Silvan Murer <silvan.murer@gmail.com>
Reviewed-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

Trivial fix to spelling mistake in module description text
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

Otherwise a race condition can occur, where userspace can start operations
before the channels have been properly initialized.
Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

Sometimes it's useful to stream samples forever, such as when
stress-testing a driver overnight to check for memory leaks or other
issues. When the program receives a signal, it will gracefully cleanup,
so it is still safe to terminate at any time.

Add support for specifying a negative -c option, meaning that we should
loop forever. To do so, we need to use a long long (instead of just
long) for num_loops so that current code specifying num_loops greater
than UNSIGNED_LONG_MAX doesn't break.
Signed-off-by: Martin Kelly <mkelly@xevo.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>