Intel Audio DSPs have "IPC" interrupt delivery and shared memory
window hardware. The Sound Open Firmware project has historically
used the combination of these to implement a bidirectional
message-passing interface. As it happens, this protocol is an
excellent fit for Zephyr's somewhat geriatric but still useful IPM
interface.
This implements a SOF-protocol-compatible transport that will
hopefully prove a bit more futureproof for non-Intel SOF
architectures. It is a software-only device, built on top of the
underlying SOC APIs for the SRAM windows (in cavs-shim) and IPC
(cavs_ipc).
Note that SOF actually has two protocol variants (ipc3 and ipc4): in
both, the command header (passed as the "id" parameter in IPM) is sent
via the hardware doorbell register. But in ipc4, the second hardware
scratch register is used to transmit the first four bytes of the
command before involving the SRAM window (in ipc3, it's ignored).
Both modes are supported by this driver, set IPM_CAVS_HOST_REGWORD to
choose the "ipc4" variant.
Finally: note that the memory layout for the windows in question is
inherited from SOF, and for compatibility (with both SOF and with the
offsets used by the host!) these can't be changed without major
surgery. They're defined in kconfig, but should be treated as
read-only until we get a chance to rework the way Zephyr does its SRAM
window management (and probably in concert with the host drivers).
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Some IPM devices support asynchronous command processing, where
acknowledgment of an IPM message can be delayed while handling
happens in a context other than the ISR that invoked the callback.
Expose this via a kconfig that can be selected by the driver, and add
a new "complete" call (a zero-overhead stub on non-supporting devices)
to signal the end of message handling.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Now that we have easy access to code on the host, it's trivial to
check the clock against host timestamps with high precision.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
On cAVS 1.8 (specifically) there seems to be a propagation delay on
the IPC registers. Hitting the TDA register to signal DONE too soon
after clearing the interrupt via TDR can cause the interrupt to be
dropped. Merely polling for it to read back correctly isn't
sufficient, we need an actual sleep here.
(The behavior that a message won't send while an existing message is
in progress is actually a hardware feature that is new with 1.8. My
guess is it's a little glitchy in its first version.)
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
On cAVS before 2.5, core power was controlled by the host. Add a
command to the cavstool.py script to allow us to do that under test
command so we can exercise multiprocessor startup/shutdown outside of
SOF.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This test obviously only works correctly if it's run from core 0 (the
only CPU that doesn't shut down). It was true essentially by
accident, but add some cpu_mask trickery to force things.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
It's useful for tests other than the IPC case to be able to
communicate with the script on the host, so generalize the interface
and move it to tests.h.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Add code to exercise soc_adsp_halt_cpu() and validate the cores can be
restarted with the kernel z_smp_start_cpu() API.
Note that this interface is for 1.8+ devices only. On cAVS 1.5, the
core power is actually controlled by the host.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Remove the soc_relaunch_cpu() API. There's now nothing that this does
that isn't already done by the kernel's own z_smp_start_cpu() API. We
don't need a SOC-level API for this anymore.
Rename soc_halt_cpu() to soc_adsp_halt_cpu() to clarify the domain,
simplify by removing the synchronization (there's nothing to
synchronize!). Also move its declaration and docs into the soc.h
header.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Things had gotten a little tangled in there so let's do some cleanup.
Remove the distressingly-special-purpose z_reinit_idle_thread() hook
(which existed to support secondary core bringup when
SMP_BOOT_DELAY=y), and just fold that into a generic z_init_cpu(),
which we can call in obvious and symmetric ways from main
initialization, z_smp_init(), and z_smp_start_cpu() (the now-official
programmatic hook for starting cpus).
Remove the "#if CONFIG_MP_NUM_CPUS > 1" exclusions. These weren't
saving any code size and were propagating themselves into platform
layers trying to avoid build failures.
There are some "special" APIs added for SOF which need to go away in
favor of the newer/generic z_smp_start_cpu(). Collect them in one
place and put them under a "#ifdef CONFIG_SOF" to prevent them from
being used in Zephyr apps.
Move some function declarations that didn't have homes into
<kernel/thread.h>.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
These are descended from a private collection of test rigs I've been
semi-curating. Getting cache, clock and memory space setup consistent
is mildly complicated on these devices and we've had bugs in all these
areas.
+ Check cache/uncache memory space setup, and make sure that the
incoherent cache is operating as expected.
+ Make sure all cores agree with clocking setup.
+ Quickly benchmark a two-instruction loop to detect messups with
instruction caching.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
As Zephyr begins to absorb drivers for these platforms that had
previously been managed by the SOF app, there's a need for a rapid
board-specific smoke test to use during development.
This starts with the smp_boot_delay test (itself a unit test for a
SOF-derived feature) and adds a host IPC case (that needs to match
code in cavstool.py on the other side of the PCI bus!).
It will grow more features over time as needed.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This is a slightly higher level Zephyr device that manages the host
IPC device for applications. There's an interface to make synchronous
and asynchronous calls, to receive commands via (interrupt context)
callbacks and emit async "done" notifications after processing is
complete. It should work for pretty much any application
architecture.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Pull latest canopennode module version. The last revision changes module
name from CANopenNode to canopennode.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
Improvement to generate Tx ack early when the event close
compared to being done at the start of the next event.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Fix Extended Advertising conditional compiles in header files.
Do not conditional compile declarations, definitions of
functions in source files only be compiled out.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Fix to include ll_adv_cmds_set when broadcast role is not
enabled in the Controller but advertising extensions is
enabled (Extended Scanning is used).
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
A previous commit fixed OpenThread logging when `LOG=n`, but
introduced regression when `LOG_MODE_MINIMAL=y`. This commit
fixes the latest.
Signed-off-by: Eduardo Montoya <eduardo.montoya@nordicsemi.no>
This commit utilizes psa_crypto_get function to fetch cryptographically
secure pseudorandom numbers.
Signed-off-by: Lukasz Duda <lukasz.duda@nordicsemi.no>
Add a new API call to replace nvs_init: nvs_mount. The new API does the
same as nvs_init except that it assumes to be provided with a valid
flash device via `struct nvs_fs` `flash_device` field. Previously, it
was not possible to avoid the runtime overhead of device_get_binding()
even if the flash device was known at compile time.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
sys_le16_to_cpu is a macro, and not big endian the value
is referenced twice, which would incorrectly increment the
value.
Modifed to extract the value to a variable before calling
sys_le16_to_cpu.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
Instead of relying on a single globally increasing (but never
resetting) packet_sequence_number, it is now correctly
reset and incremented for each ISO channel.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
Adding the same BT callback struct twice can cause
an infinite loop when iterating the callback_list,
so register the gatt callback just once early.
Signed-off-by: Peter Johanson <peter@peterjohanson.com>
Differentiate integration)platforms list per each sample test.
Removed sample.mcumg.smp_svr.bt_tiny as it can not fit in
nr51422 flash anymore, even with configuration tweeks.
Signed-off-by: Andrzej Puzdrowski <andrzej.puzdrowski@nordicsemi.no>
Rename the use of mstr word in advanced scheduling
implementation to cen as a representation for Central
role.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Use connection interval units of 1250 us for periodic
scheduling of Extended Advertising auxiliary PDU events so
that auxiliary events can be periodically grouped alongwith
the Periodic Advertising Events. This will permit
mitigating overlaps amongst them.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Add implementation to find free slot after Periodic
Advertising event, to place the BIG event.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Make the Periodic Advertising time reservation calculation
function internal public so that advanced scheduling can
use it to scheduling the BIG event after the Periodic
Advertising radio events.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
Refactor the implementation to get free slot after a
state/role so that it is easy for reused for scheduling
Broadcast ISO events.
Signed-off-by: Vinayak Kariappa Chettimada <vich@nordicsemi.no>
