The Cascada 8210 hardware transceiver is kind of a hardMAC which
interfaces with the softMAC and in practice does not support sending
anything else than dataframes. This means we cannot send any BEACON_REQ
during active scans nor any BEACON in general. Refuse these operations
officially so that the user is aware of the limitation.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Let's create a couple of driver hooks in order to tell the device
drivers that beacons are being sent, allowing them to eventually apply
any needed configuration, or in the worst case to refuse the operation
if it is not supported. These hooks are optional and not implementing
them does not prevent the beacons operation to happen. Returning an
error from these implementations will however shut down the beacons
configuration entirely.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Let's create a couple of driver hooks in order to tell the device
drivers that a scan is ongoing, allowing them to eventually apply any
needed configuration, or in the worst case to refuse the operation if it
is not supported. These hooks are optional and not implementing them
does not prevent the scan operation to happen. Returning an error from
these implementations will however shut down the scan entirely.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

When performing an active scan, coordinators emit beacon requests which
must be answered by other PANs receiving the request.

Answering a beacon request is considered a duty whenever the user
performs a "send beacons" command with an interval of 15. As in the
softMAC we save the interval in micro-seconds, we use a negative value
to discriminate between a passive beacons command and the task to answer
received beacon requests.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Active scan support is based on the current passive scan support,
cheered up with beacon requests sent after every channel change.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Add an internal trace when new PANs get discovered.

Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Implement the core hooks in order to provide the softMAC layer support
for sending beacons. Besides being able to test the full passive
scanning procedure, this will also be used when defining PAN coordinator
status, in order to give a device the right to answer received
BEACON_REQ.

Changing the channels is prohibited while a beacon operation is
ongoing.

The implementation uses a workqueue triggered at a certain interval
depending on the symbol duration for the current channel and the
interval order provided.

Sending beacons in response to an active scan request is not
yet supported.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

This involves processing beacons requests: starting or stopping
the flow of beacons sent passively on a specific interface. The page and
channel must be changed beforehands if needed. Interval orders above 14
are reserved to tell a device it must answer BEACON_REQ coming from an
active scan procedure (this is not supported yet).

A beacons request structure is created to list the requirements.

Mac layers may now implement the ->send_beacons() and
->stop_beacons() hooks.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Now that scanning is supported and PANs properly registered, give
certain rights to the user, such as listing asynchronously the listed
PANs as well as flushing the list.

The maximum number of PANs to list and their delay before expiration can
be configured. By default there is no limit. When these parameters are
set, PANs are automatically dropped from the list.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Implement the core hooks in order to provide the softMAC layer support
for scan requests and aborts.

Changing the channels is prohibited during the scan.

As transceiver enter in a promiscuous mode during scans, they might stop
checking frame validity so we ensure this gets done at mac level.

The implementation uses a workqueue triggered at a certain interval
depending on the symbol duration for the current channel and the
duration order provided.

Received beacons during a passive scanning procedure are processed and
either registered in the list of known PANs or the existing entry gets
updated accordingly.

Active scanning is not supported yet.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

This involves processing triggering scan requests, abort scan requests,
as well as providing information about if the last scan was finished or
not.

A scan request structure is created to list the requirements.

A netlink multicast scan group is also created for the occasion so that
listeners can only focus on scan activity.

Mac layers may now implement the ->trigger_scan() and ->abort_scan()
hooks.

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

A 802.15.4 frame can be of different types, here is a definition
matching the specification. This enumeration will be soon be used when
adding scanning support.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

This definition will be used when adding support for scanning and defines
the content of a beacon frame header as in the 802.15.4 specification.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Let's introduce the basics of PAN management:
- structures defining PANs
- helpers for PANs registration
- helpers discarding old PANs

Co-developed-by: David Girault <david.girault@qorvo.com>
Signed-off-by: David Girault <david.girault@qorvo.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Returning -1 does not indicate anything useful.

Use a standard and meaningful error code instead.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

…ture

While not being absolutely needed, it at least explain the mode vs. enum
fields.

Signed-off-by: David Girault <david.girault@qorvo.com>
[miquel.raynal@bootlin.com: Isolate this change from a bigger commit and
                            reword the comment]
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Move the address structure earlier in the cfg802154.h header in order to
use it in subsequent additions. There is no functional change here.

Signed-off-by: David Girault <david.girault@qorvo.com>
[miquel.raynal@bootlin.com: Isolate this change from a bigger commit]
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

The softMAC stack has no meaning outside of the IEEE 802.15.4 stack and
cannot be used without it.

Signed-off-by: David Girault <david.girault@qorvo.com>
[miquel.raynal@bootlin.com: Isolate this change from a bigger commit]
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

It makes certainly more sense to have all the low-range wireless
protocols such as Bluetooth, IEEE 802.11 (WiFi) and IEEE 802.15.4
together, so let's move the main IEEE 802.15.4 stack Kconfig entry at a
better location.

Signed-off-by: David Girault <david.girault@qorvo.com>
[miquel.raynal@bootlin.com: Isolate this change from a bigger commit and
rewrite the commit message.]
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

…already

The core now knows how to set the symbol duration in a few cases, when
drivers correctly advertise the protocols used on each channel. For
these drivers, there is no more need to bother with symbol duration, so
just drop the duplicated code.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

There is no point in accepting frames with a wrong or missing checksum,
at least not outside of a promiscuous setting. Set the right flag by
default in the hwsim driver to ensure checksums are not ignored.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Now that we have access to all the basic information to know which
symbol duration should be applied, let's set the symbol duration
automatically. The two locations that must call for the symbol duration
to be set are:
- when manually requesting a channel change though the netlink interface
- at PHY creation, ieee802154_alloc_hw() already calls
  ieee802154_change_channel() which will now update the symbol duration
  accordingly.

If an information is missing, the symbol duration is not touched, a
debug message is eventually printed. This keeps the compatibility with
the unconverted drivers for which it was too complicated for me to find
their precise information. If they initially provided a symbol duration,
it would be kept. If they don't, the symbol duration value is left
untouched.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Tdsym is often given in the spec as pretty small numbers in microseconds
and hence was reflected in the code as symbol_duration and was stored as
a u8. Actually, for UWB PHYs, the symbol duration is given in
nanoseconds and are as precise as picoseconds. In order to handle better
these PHYs, change the type of symbol_duration to u32 and store this
value in nanoseconds.

All the users of this variable are updated in a mechanical change.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

…igurations

In order to let the core derive eg. the symbol duration for a given
channel, it needs to know which protocol is being used, on which band,
and eventually more details such as the mean PRF in the case of UWB.

Create the necessary enumerations to declare all of that. Include them
in the currently-almost-empty phy_channels structure which until now
only declared the supported channels as bitfields.

The PRF is declared in a union as this clearly is a parameter that does
not apply to most of the protocols. It is very likely that other
parameters will get added in the future to further define specific
protocols and the union will likely be a good location for them.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

The idea here is to create a structure per set of channels so that we
can define much more than basic bitfields for these.

The structure is currently almost empty on purpose because this change
is supposed to be a mechanical update without additional information but
more details will be added in the following commits.

For each page, the core now has access to how many "chunks" of channels
are defined. Overall up to only 3 chunks have been defined so let's
hardcode this value to simplify a lot the handling. Then, for each
chunk, we define an independent bitfield of channels. As there are
several users of these bitfields, we also create the
cfg802154_get_supported_chans() helper to reconstruct the bitfield as it
was before when the only information that matters is identifying the
supported/unsupported channels.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

Right now device drivers are encouraged to just set the ->current_page
and ->current_channel to indicate about the current state of the
hardware but this is far from ideal given the fact that we might want to
configure a few things internally, such as the symbol duration.

Call the ieee802154_change_channel() helper from the code section
registering the hardware to ensure proper channel selection and
configuration both on the device side and the core side.

This change somehow "fixes" the hwsim driver which advertises using page
0 channel 13, but does not actually update its own internal pib
structure to reflect that configuration.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

As it is currently designed, the set_channel() cfg802154 hook
implemented in the softMAC is doing a couple of checks before actually
performing the channel change. However, as we enhance the support for
automatically setting the symbol duration during channel changes, it
will also be needed to ensure that the corresponding channel as properly
be selected at probe time. In order to verify this, we will need to
separate the channel change from the previous checks. These checks are
actually here to speed up the process when the user request necessitates
no change, so we can safely bypass them.

Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>

…m/linux/kernel/git/tip/tip

Pull locking updates from Borislav Petkov:
 "Lots of cleanups and preparation. Highlights:

   - futex: Cleanup and remove runtime futex_cmpxchg detection

   - rtmutex: Some fixes for the PREEMPT_RT locking infrastructure

   - kcsan: Share owner_on_cpu() between mutex,rtmutex and rwsem and
     annotate the racy owner->on_cpu access *once*.

   - atomic64: Dead-Code-Elemination"

[ Description above by Peter Zijlstra ]

* tag 'locking_core_for_v5.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/atomic: atomic64: Remove unusable atomic ops
  futex: Fix additional regressions
  locking: Allow to include asm/spinlock_types.h from linux/spinlock_types_raw.h
  x86/mm: Include spinlock_t definition in pgtable.
  locking: Mark racy reads of owner->on_cpu
  locking: Make owner_on_cpu() into <linux/sched.h>
  lockdep/selftests: Adapt ww-tests for PREEMPT_RT
  lockdep/selftests: Skip the softirq related tests on PREEMPT_RT
  lockdep/selftests: Unbalanced migrate_disable() & rcu_read_lock().
  lockdep/selftests: Avoid using local_lock_{acquire|release}().
  lockdep: Remove softirq accounting on PREEMPT_RT.
  locking/rtmutex: Add rt_mutex_lock_nest_lock() and rt_mutex_lock_killable().
  locking/rtmutex: Squash self-deadlock check for ww_rt_mutex.
  locking: Remove rt_rwlock_is_contended().
  sched: Trigger warning if ->migration_disabled counter underflows.
  futex: Fix sparc32/m68k/nds32 build regression
  futex: Remove futex_cmpxchg detection
  futex: Ensure futex_atomic_cmpxchg_inatomic() is present
  kernel/locking: Use a pointer in ww_mutex_trylock().

…linux/kernel/git/tip/tip

Pull scheduler updates from Borislav Petkov:
 "Mostly minor things this time; some highlights:

   - core-sched: Add 'Forced Idle' accounting; this allows to track how
     much CPU time is 'lost' due to core scheduling constraints.

   - psi: Fix for MEM_FULL; a task running reclaim would be counted as a
     runnable task and prevent MEM_FULL from being reported.

   - cpuacct: Long standing fixes for some cgroup accounting issues.

   - rt: Bandwidth timer could, under unusual circumstances, be failed
     to armed, leading to indefinite throttling."

[ Description above by Peter Zijlstra ]

* tag 'sched_core_for_v5.17_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/fair: Replace CFS internal cpu_util() with cpu_util_cfs()
  sched/fair: Cleanup task_util and capacity type
  sched/rt: Try to restart rt period timer when rt runtime exceeded
  sched/fair: Document the slow path and fast path in select_task_rq_fair
  sched/fair: Fix per-CPU kthread and wakee stacking for asym CPU capacity
  sched/fair: Fix detection of per-CPU kthreads waking a task
  sched/cpuacct: Make user/system times in cpuacct.stat more precise
  sched/cpuacct: Fix user/system in shown cpuacct.usage*
  cpuacct: Convert BUG_ON() to WARN_ON_ONCE()
  cputime, cpuacct: Include guest time in user time in cpuacct.stat
  psi: Fix PSI_MEM_FULL state when tasks are in memstall and doing reclaim
  sched/core: Forced idle accounting
  psi: Add a missing SPDX license header
  psi: Remove repeated verbose comment

…scm/linux/kernel/git/gustavoars/linux

Pull cast-function-type warning addition from Gustavo A. R. Silva:
 "This globally enables -Wcast-function-type.

  We need to make sure new function cast mismatches are not introduced
  in the kernel in order to avoid tripping CFI checking"

* tag 'Wcast-function-type-5.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gustavoars/linux:
  Makefile: Enable -Wcast-function-type

Pull ARM updates from Russell King:

 - amba bus irq rework

 - add kfence support

 - support for Cortex M33 and M55 CPUs

 - kbuild updates for decompressor

 - let core code manage thread_info::cpu

 - avoid unpredictable NOP encoding in decompressor

 - reduce information printed in calltraces

* tag 'for-linus' of git://git.armlinux.org.uk/~rmk/linux-arm:
  ARM: reduce the information printed in call traces
  ARM: 9168/1: Add support for Cortex-M55 processor
  ARM: 9167/1: Add support for Cortex-M33 processor
  ARM: 9166/1: Support KFENCE for ARM
  ARM: 9165/1: mm: Provide is_write_fault()
  ARM: 9164/1: mm: Provide set_memory_valid()
  ARM: 9163/1: amba: Move of_amba_device_decode_irq() into amba_probe()
  ARM: 9162/1: amba: Kill sysfs attribute file of irq
  ARM: 9161/1: mm: mark private VM_FAULT_X defines as vm_fault_t
  ARM: 9159/1: decompressor: Avoid UNPREDICTABLE NOP encoding
  ARM: 9158/1: leave it to core code to manage thread_info::cpu
  ARM: 9154/1: decompressor: do not copy source files while building

…ernel/git/deller/parisc-linux

Pull parisc architecture updates from Helge Deller:

 - Fix lpa and lpa_user defines (John David Anglin)

 - Fix symbol lookup of init functions with an __is_kernel() fix (Helge
   Deller)

 - Fix wrong pdc_toc_pim_11 and pdc_toc_pim_20 definitions (Helge
   Deller)

 - Add lws_atomic_xchg and lws_atomic_store syscalls (John David Anglin)

 - Rewrite light-weight syscall and futex code (John David Anglin)

 - Enable TOC (transfer of contents) feature unconditionally (Helge
   Deller)

 - Improve fault handler messages (John David Anglin)

 - Improve build process (Masahiro Yamada)

 - Reduce kernel code footprint of user access functions (Helge Deller)

 - Fix build error due to outX() macros (Bart Van Assche)

 - Ue default_groups in kobj_type in pdc_stable (Greg Kroah-Hartman)

 - Default to 16 CPUs on 32-bit kernel (Helge Deller)

* tag 'for-5.17/parisc-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Default to 16 CPUs on 32-bit kernel
  sections: Fix __is_kernel() to include init ranges
  parisc: Re-use toc_stack as hpmc_stack
  parisc: Enable TOC (transfer of contents) feature unconditionally
  parisc: io: Improve the outb(), outw() and outl() macros
  parisc: pdc_stable: use default_groups in kobj_type
  parisc: Add kgdb io_module to read chars via PDC
  parisc: Fix pdc_toc_pim_11 and pdc_toc_pim_20 definitions
  parisc: Add lws_atomic_xchg and lws_atomic_store syscalls
  parisc: Rewrite light-weight syscall and futex code
  parisc: Enhance page fault termination message
  parisc: Don't call faulthandler_disabled() in do_page_fault()
  parisc: Switch user access functions to signal errors in r29 instead of r8
  parisc: Avoid calling faulthandler_disabled() twice
  parisc: Fix lpa and lpa_user defines
  parisc: Define depi_safe macro
  parisc: decompressor: do not copy source files while building

…m/linux/kernel/git/tip/tip

Pull missed x86 build updates from Borislav Petkov:
 "In the heat of the battle between me and the tip pull requests, I've
  sent you the tip:x86/build branch without the two top-most commits
  from Ingo (it appears I hadn't fast-forwarded my local branch, sorry
  about that).

  Therefore, here's a part 2 of x86/build for 5.17.

   - Make x86 defconfigs more useful by enabling
     CONFIG_LOCALVERSION_AUTO and CONFIG_KALLSYMS_ALL

  And while you merge that, I'll be at the blackboard, writing:

    I will always fast-forward local branches before tagging and sending pull requests to Linus.
    I will always fast-forward local branches before tagging and sending pull requests to Linus.
    I will always fast-forward local branches before tagging and sending pull requests..."

* tag 'x86_build_for_v5.17_rc1-p2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/kbuild: Enable CONFIG_KALLSYMS_ALL=y in the defconfigs
  x86/defconfig: Enable CONFIG_LOCALVERSION_AUTO=y in the defconfig

…kernel/git/efi/efi

Pull EFI updates from Ard Biesheuvel:

 - support taking the measurement of the initrd when loaded via the
   LoadFile2 protocol

 - kobject API cleanup from Greg

 - some header file whitespace fixes

* tag 'efi-next-for-v5.17' of git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi:
  efi: use default_groups in kobj_type
  efi/libstub: measure loaded initrd info into the TPM
  efi/libstub: consolidate initrd handling across architectures
  efi/libstub: x86/mixed: increase supported argument count
  efi/libstub: add prototype of efi_tcg2_protocol::hash_log_extend_event()
  include/linux/efi.h: Remove unneeded whitespaces before tabs

…kernel/git/rw/uml

Pull UML updates from Richard Weinberger:

 - set_fs removal

 - Devicetree support

 - Many cleanups from Al

 - Various virtio and build related fixes

* tag 'for-linus-5.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/uml: (31 commits)
  um: virtio_uml: Allow probing from devicetree
  um: Add devicetree support
  um: Extract load file helper from initrd.c
  um: remove set_fs
  hostfs: Fix writeback of dirty pages
  um: Use swap() to make code cleaner
  um: header debriding - sigio.h
  um: header debriding - os.h
  um: header debriding - net_*.h
  um: header debriding - mem_user.h
  um: header debriding - activate_ipi()
  um: common-offsets.h debriding...
  um, x86: bury crypto_tfm_ctx_offset
  um: unexport handle_page_fault()
  um: remove a dangling extern of syscall_trace()
  um: kill unused cpu()
  uml/i386: missing include in barrier.h
  um: stop polluting the namespace with registers.h contents
  logic_io instance of iounmap() needs volatile on argument
  um: move amd64 variant of mmap(2) to arch/x86/um/syscalls_64.c
  ...