This sample builds multiple kernels with different configurations
starting with a minimal kernel that basically does nothing and
adds features one by one to measure the minimal footprint for the
various supported kernel objects and features.
Change-Id: Icba984845dc6f1979d16ba246bdb133cc6e3465c
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The event portion of the app_kernel benchmark no longer outputs the following
(abbreviated) error message:
<path to project>/app_kernel/src/event_b.c:86 Error: tick occurred
This error was occurring as the benchmark test had hard-coded the value of
the event TEST_EVENT to 0 instead of defining it in the prj.mdef file.
Consequently, the system was using event 0 for the tick event; every time
task_event_send(TEST_EVENT) was invoked, the system treated it as a tick event.
Change-Id: I1a785a6594415bd4b0300d382f76a6d768641864
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Other apps are being compile-tested, do this here too.
Change-Id: Ibbe9e765757564e974233c1410bac320ae7a8a68
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The profiler now uses the generic ring buffer. The dropped event
count is stored in the ring buffer's value field. The data size only
refers to the extra data attached to the message and NOT any internal
representation of metadata inside the ring buffer, the event_logger
APIs now pass this information along in dedicated parameters.
Change-Id: I1f168e6a05e8d937bf86b2a4cccecbb04b0118c6
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This patch is based on some code contributed by Dirk Brandewie.
This is a generic data structure for queuing data in a FIFO ring
buffer. Any given data enqueued is annotated with an app-specific
type identifier, and small integral value. Use of a data pointer
is optional if the necessary information can be conveyed in the
annotations. We want all the metadata to fit in a single DWORD.
The ring buffer always contains at least one free dword in the
buffer to correctly distinguish between full and empty queues.
Concurrency control is almost nonexistent; depending on usage,
apps may want to introduce the usage of semaphores and/or mutexes
to preserve the integrity of the ring buffer or provide notifications
when data is available.
Change-Id: I860262d2afc96db4476d4c695a92f7da355ab732
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Add interrupt and sleep event data to the profiler sample.
Change-Id: Ie5d3c4201475824eaf833bef506140279a6c606d
Signed-off-by: Yonattan Louise <yonattan.a.louise.mendoza@intel.com>
Removed the term _node from the document as requested.
Change-Id: I11e7f9fb217d6cc1026e81fb4702fa534de53874
Signed-off-by: Gerardo A. Aceves <gerardo.aceves@intel.com>
CONF_FILE was dropped leaving samples build with
default options.
Change-Id: Id1a3cb06f43052fb74ed9edcc2ea275b44d2b6b4
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Make these public:
- SECONDS(x): macro that gives the number of ticks in x seconds
- MSEC(x): macro that gives the number of ticks in x milliseconds
- MSEC_PER_SEC: number of milliseconds per second
- USEC_PER_MSEC: number of microseconds per millisecond
Change-Id: Ic5dbf9349651a477b066edb0c6b6721da2b7e5bb
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
The term 'context' is vague and overloaded. Its usage for 'an execution
context' is now referred as such, in both comments and some APIs' names.
When the execution context can only be a fiber or a task (i.e. not an
ISR), it is referred to as a 'thread', again in comments and everywhere
in the code.
APIs that had their names changed:
- nano_context_id_t is now nano_thread_id_t
- context_self_get() is now sys_thread_self_get()
- context_type_get() is now sys_execution_context_type_get()
- context_custom_data_set/get() are now
sys_thread_custom_data_set/get()
The 'context' prefix namespace does not have to be reserved by the
kernel anymore.
The Context Control Structure (CCS) data structure is now the Thread
Control Structure (TCS):
- struct ccs is now struct tcs
- tCCS is now tTCS
Change-Id: I7526a76c5b01e7c86333078e2d2e77c9feef5364
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
The comment did not convey anything, as it was a relic from before a
previous renaming. Use the SECONDS(x) macro to make everything clearer.
Change-Id: Ia757061c4083d7567df5b214326c2cf8b6804fbf
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
This is just a cosmetic change to unify anything interrupt related
under 'irq'.
Change-Id: Ib8804d194e11eb49526fda952d9efc0f2ffac2df
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Previously, CONFIG_PRINTK being false means none of functions related to
the UART console and serial are compiled into the binary. This
effectively means UART console and serial drivers are disabled.
By decoupling serial drivers from console related configs, additional
options are needed to disable serial drivers and UART console driver
itself. So add those options here to disable UART console and serial,
or else the resulting binaries will be too bloated.
Change-Id: If526e42404f22caf6a550795f8277ba742625883
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This restructures the device structs and now utilizes the new
driver initialization model. This is another step towards
converting the serial driver to the new driver model.
Note that the serial driver does not initialize the hardware
unless it is being used by another driver. The configuration
of the serial port needs to be done by the driver utilizing
the port (e.g. baud rate, interrupt priority, etc.).
Therefore, some serial ports are declared but not exactly
configured.
Also note that the UART console is being initialized at
the same time as the serial port. This will be removed
in future patch, so the UART console driver will do
its own initialization.
Change-Id: Idd89954b2d0649a557ba8c869ee96512fec898e4
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
... instead of an array index to a global array. This is
an intermediate step to make the drivers conform to
the new driver model.
This only changes from using a direct array index to using
device structs. The UARTs are still staticlly defined.
Later patches will make the drivers utilize the driver
initialization procedure specified by the driver model.
Change-Id: I18041bbb4b0efdf8ae87088fd000b391d0827e9b
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
The test was doing some weird dance for computing the time the main task
had slept, context switching between the main task and a helper task.
The tick timestamp that marks the start of the sleep duration, which
should be taken before the main task goes to sleep actually was taken
after it went to sleep, by the helper task. This lead to weird results
sometimes on QEMU, where the main task would report that it had slept
for less time than requested.
Now instead, the main task takes its own tick timestamps, just before it
goes to sleep and right when it awakens.
The helper task takes a timestamp as well, to verify that it did do busy
work while the main task was sleeping. However, some error is allowed
there to compensate for some QEMU weirdness.
Change-Id: I4b642b49de8346be404000698eaa4ded070d4097
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
The build system is actually using the output directory
as the working dir for the Make session, using a symlink
to the source directory. Relative paths don't work correctly,
it only works now due to other issues in the build system
where absolute paths are being used instead for app source
files (causing other problems).
Change-Id: I2bcd82314692902f12da51c96fe912efb68bdc5e
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Instead of using a relative path in an #include, use a real -I path
in CFLAGS. This is still a hack, what we really need is support in
the build system for generating static libraries.
Change-Id: I7bc9aba3bcb7ec65f6a6119c3b2c6efc1de67e58
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Bluetooth shell now supports Signed Write if CSRK key is present.
Change-Id: I035a0d314da4997105c64da58cf85bd95fdbaa5b
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Add support for signed write with sign parameter for write without
response.
Change-Id: I79008532d88b10d34db1f68898ad4258dd3e761b
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
This allows to register pairing callbacks that will display passkey
in case of authenticated pairing.
Change-Id: I18422193785fdcd8a2e567adeb65199bfadcdd1f
Signed-off-by: Szymon Janc <ext.szymon.janc@tieto.com>
The DTLS server is running in Zephyr and is listening on port 4242
and when it receives DTLS UDP packet, it will reverse the data and
send it back to the caller.
Use the dtls-client application running in host to connect to
the dtls-server.
Change-Id: I3306cb242498019ffc9e0519b06061c36db96f78
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Make stack size the same size as in hello app. Current stack cannot
handle correctly security tests.
Change-Id: I5946c6bbcbc5079cefefddbb229b3205364b24d0
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
This adds gatt-write-without-response which can be used as following:
btshell> gatt-write-without-response <handle> <data>
bt: bt_gatt_write_without_response (0x0010f0a0): handle 0x0011 length 1
Write Complete (err 0)
Change-Id: Ide166049cf53e691a2269778bca1489f7027621f
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Now uses net_receive() with a proper timeout. Removed also
extra waits when sending data. Removed extra semaphores
and streamlined the code to act as fast as possible.
Change-Id: I577d4d5ff85bf13400fb6e157bc0023267658ba4
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The echo-client app will run in Zephyr and will send network
packets to host process (echo-server) via SLIP.
Change-Id: Icaf941ce883aec82136ef909db67e43ab6b981f7
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
We must not touch uip_len(buf) value as Contiki IP stack
will use it to determine what it should do with the
received packet. So after we have received the packet,
we update separate datalen variable in net_buf that will
store the correct length of the packet.
Change-Id: Iab79b741508e95e581d1727645e6b1d1eacded4c
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Allow caller to specify a timeout in ticks while waiting
network data. The timeout value has following meaning:
TICKS_UNLIMITED wait forever
TICKS_NONE do not wait
value > 0 timeout value (only applicaple if
CONFIG_NANO_TIMEOUTS is defined)
Change-Id: Icbb33fa0eeb462659d4006dd9d948a04f39a87a9
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
This adds offset to gatt-write which can be used as following:
btshell> gatt-write 0011 0001 01
bt: gatt_prepare_write (0x0010f680): handle 0x0011 offset 1 len 1
Write pending
btshell> bt: bt_att_recv (0x00111918): Received ATT code 0x17 len 6
bt: att_handle_prepare_write_rsp (0x00111918):
bt: att_prepare_write_rsp (0x00111918): err 0x00
bt: gatt_exec_write (0x00111918):
bt: bt_att_recv (0x00111918): Received ATT code 0x19 len 1
bt: att_handle_exec_write_rsp (0x00111918):
bt: att_write_rsp (0x00111918): err 0x00
Write complete: err 0
Change-Id: I4d3c038ea572da92efe0de8ce426ffca15506440
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
This enables writing from offset using Prepare Write + Execute Write
procedures including Long Write procedure when the data is bigger than
the MTU.
Change-Id: Id35d66c011ba582a4ba92ffe75444226e479335a
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Sample project README.txt files are supposed to follow a particular format
AND have a "Sample Output:" section.
Change-Id: I91fff1057e56bc222ab82d680018a673e19e479d
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Assigns more descriptive titles to the sample project README.txt files.
Change-Id: I4a0efdac481a8ed75a847871dbfe686e74a9f113
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Standardize README.txt files on using "Sample Output:".
Change-Id: I3a8dc18b37f36067b9f343ef443b6d35e85db6b4
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Standardize on using the irq_lock/irq_unlock (non-inline) symbols
everywhere.
The non-inline versions provide absolutely no benefits, so they will be
removed in a subsequent commit, and the inline versions will have their
_inline suffix removed.
Change-Id: Ib0b55f450447366468723e065a60adbadf7067a9
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
Basically, this does not work anymore:
task_timer_start(<timer>, 0, X, <sem>);
since it does not make much sense to have a timer with an expiry of 0
ticks. The code internally was setting the duration to be equal to the
period anyway. So, to achieve the same behaviour, do this:
task_timer_start(<timer>, X, X, <sem>);
This has the positive side-effect of removing a small block of logic
that was handling the cases where duration was 0.
Change-Id: Ic4af4a17a129f14af4fea445bcaddabe89c27131
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
Reflect the fact that they are not used.
Change-Id: I6ea83cb2c7532e10988cbf4350edccb78f444328
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
Align with the newer terminology used for microkernel internal symbols.
Change-Id: I623b383f90d9e37a49429a79774c7f7a4953bd5f
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
irq_handler_set, irq_priority_set and irq_disconnect have been made
private by prepending an underscore to their names:
irq_handler_set -> irq_handler_set
irq_priority_set -> irq_priority_set
irq_disconnect -> irq_disconnect
The prototypes have been removed from header files when possible, and
extern statements used in C code where they were called.
_irq_priority_set() for ARM is still in the header file because
IRQ_CONFIG() relies on it.
Change-Id: I2ad585f8156ff80250f6d9eeca4a249a4477fd9d
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
Doing those checks went against the Zephyr philosophy of no error
checking unless absolutely necessary. Users should ensure themselves the
validity of their inputs to kernel APIs.
Change-Id: I21e5cd07ff9424ad61e81fd9d52ceef0c9584a8c
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
task_event_set_handler -> task_event_handler_set
Align with the "verb at the end" convention.
Change-Id: I8b72d41a20a7fdd4756f90765682e317289a241b
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
task_pipe_put_async -> task_pipe_block_put
task_mbox_put_async -> task_mbox_block_put
task_mbox_data_get_async_block -> task_mbox_data_block_get
task_mbox_data_get_async_block_wait -> task_mbox_data_block_get_wait
task_mbox_data_get_async_block_wait_timeout ->
task_mbox_data_block_get_wait_timeout
Previous names, focusing on 'async', were misleading, because:
- some of those APIs can be used synchronously as well
- other APIs can also do asynchronous transfer, and don't have 'async'
in their names
- the key concept of these APIs is that they use memory pool blocks
rather than raw data buffers.
Change-Id: I0c08a6cf950ab23bb4172ce25eb6f9886b037649
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
Add a project that shows how to collect the
data from the profiler.
Change-Id: Ie9cc88290de594720739b30d4e872fe51d353f6e
Signed-off-by: Yonattan Louise <yonattan.a.louise.mendoza@intel.com>
leftovers from the renaming activities:
nano_node_tick_get_32 -> nano_tick_get_32
Change-Id: Ic9f0870ddda159311005c2ed16e0a378ab9a6a62
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The PIT is an unsupported device. On x86 based platforms, the LOAPIC_TIMER and
HPET_TIMER are the only supported timers.
Change-Id: Ic890838c811b7eb62008aef0c8a92786f1579217
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Removes the LOAPIC timer macros from board.h as they are no longer required
since appropriate Kconfig options exist to use in their stead.
Change-Id: Id0b509eba7cce1839b400f95b989bf886322e1a8
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Anas Nashif