k_cycle_get_32() needs to return a monotonically increasing value,
except in cases of 32-bit integer overflow. Enforce this with a
test case.
We also check that the number of cycles elapsed after sleeping for 1
second is at the expected value. This can help catch errors on platforms
that use different timer sources for the system clock and timestamps.
This test case adapted from some code provided by Sergey Arkhipov
when troubleshooting ZEP-1546.
Issue: ZEP-1546
Change-Id: If27fff026ea6de659f7b41b60ff26f4962b734d4
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The LOAPIC driver was doing this in a way susceptible to a very
nasty race condition: the CCR register could reset and be readable
before the associated interrupt could be delivered.
This resulted in a small window of time where CCR was reset, but
accumulated_cycle_count not updated, causing some calls to
k_cycle_get_32() to appear to jump backwards in time.
Just use the x86 TSC for these cycle timestamps. A divisor may be
provided in cases where the CPU clock speed is some multiple of
the bus speed. Modern x86 CPUs do not change their TSC rate even
when adjusting cpu frequency, so this should be a reliable timing
source.
Issue: ZEP-1546
Change-Id: I441bd8e32af866587a91f306e89e3fa0ece512b5
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
It's possible the timer interrupt could occur when performing the
computation, resulting in incorrect values returned.
It's still possible for bad values to be returned if the function is
called with interrupts locked, but that is only fixable with a second
timer source.
Issue: ZEP-1546
Change-Id: I16d5b04c3e32377f7249eb4fb1bf2f7c22bd0836
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Some files made it through review process with full license header.
Change-Id: I2722b127c40b4b19500042c12e4fde85a165bae9
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Put them in order where they are most likely to be useful.
Change-Id: Ia9c358a096556a9838b2b69311e10aba3b9ca587
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Commit 7cb8a16c86 ("doc: restructure application primer") removed
the section documenting the build system support for third party
libraries. Restore this section making a few editorial changes to
ensure the text sits well in its surroundings.
Jira: ZEP-1733
Change-Id: Ie62b956732f36fac70b392eeee880acebaef6cf9
Signed-off-by: Daniel Thompson <daniel.thompson@linaro.org>
Promote a glossary.rst up into the doc/ folder, merge wiki
glossary entries (and remove from the wiki), and format use
the .. glossary directive to allow references by using the
:term: role (using :term:`ISR` will make a link to the
glossary entry for ISR)
Jira: ZEP-1321
Change-Id: Ie1461037ab456371604594488f01df9f21284561
Signed-off-by: David B. Kinder <david.b.kinder@intel.com>
This patch adds the UART ports required by for Bluetooth. Baud rate
was moved from the Kinetis dtsi file to the relevant board files.
Jira: ZEP-1745
Change-Id: Iac4f748fd82217662800dbf48baea087e5d3a1df
Signed-off-by: Andy Gross <andy.gross@linaro.org>
This section was renamed to kinetis_flash_config, but it slipped back in
with the previous section_tags.h cleanup.
Change-Id: I4a2558aa70379a75c604f3617483680a35ed3d90
Signed-off-by: Maureen Helm <maureen.helm@nxp.com>
The isalnum() primitive is used by the NATS protocol implementation to
vaildate some of the inputs.
This uses primitives that were already in place.
Change-Id: Ib53eeb7ae002a42f5b6aa8d4fc61baca029a042d
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
This is a reworked version of the previous RTC driver. The main
changed is related to the handling _timer_idle_exit() on non-RTC
wake-ups. The previous version didn't announce the elapsed time
to the kernel in _timer_idle_exit(). Additionally, the driver now
makes sure never to announce more idle ticks than the kernel asked
for, since the kernel does not handle negative deltas in its timeout
queues.
Change-Id: I312a357a7ce8f0c22adf5153731064b92870e47e
Signed-off-by: Wojciech Bober <wojciech.bober@nordicsemi.no>
Signed-off-by: Øyvind Hovdsveen <oyvind.hovdsveen@nordicsemi.no>
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Some arches may want to define this as an inline function, or
define in core arch code instead of timer driver code.
Unfortunately, this means we need to remove from the footprint
tests, but this is not typically a large function.
Issue: ZEP-1546
Change-Id: Ic0d7a33507da855995838f4703d872cd613a2ca2
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
On Nios II the same timer peripheral IP block can't function
as a periodic system timer and a high-resolution timestamp source.
A second timer instance with different configuration is required.
Until that is implemented, just return the accumulated cycle count.
Issue: ZEP-1546
Change-Id: If3dcebdc60334bf3aa0ab45ccd82f1b2531b6bc1
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
There are race conditions trying to coordinate the value between
the accumulated_cycle_count (updated at interrupt time) and
trying to compute the delta from the last interrupt using the
mtime registers. An unlucky call could result in the timestamp
appearing to move backwards in time.
the 'mtime' register isn't reset at every interrupt. Since we just
want a cycle counter, report its raw value.
Issue: ZEP-1546
Change-Id: I9f404b33214d6502fea47374fcf0ecbf84ef8136
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Currently, if IRQ number != RISCV_MACHINE_TIMER_IRQ (only device IRQ in qemu),
riscv32-qemu was considering the IRQ as an exception. However, fake IRQs
can also be generated by setting corresponding bits in the Machine Interrupt
Pending register (mip). With the current implementation, these IRQs were
considered as unexpected exceptions.
To circumvent the problem, update the IRQ filtering mechanism by considering
an IRQ (IRQ number as reported by the mcause register) as an exception only
if its corresponding bit is NOT set in the mip register.
Change-Id: I4c581a84d83ee0ba2c4ea35f89ba732401eb8fa4
Signed-off-by: Jean-Paul Etienne <fractalclone@gmail.com>
This driver currently only handles memory-to-memory transfers.
TODO: Add support for all types of transfers.
Change-Id: Ic9787dcca919a58fb11b48c9f6b6b371db88c3ea
Origin: Original
Maintained-by: Zephyr
Signed-off-by: Lee Jones <lee.jones@linaro.org>
A _tsc_read has been added for ARC targets.
This test can use that when ARC.
See ZEP-1413
Change-Id: Ib63aecbe9f3eb2b97ad1086fc79b57e8f0774fca
Signed-off-by: Chuck Jordan <cjordan@synopsys.com>
Some inconsistent spacing and private types starting with '_'.
Change-Id: I3354b69cc3934717d3b8097cdda98474339c1f32
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
The loop was not tracking the correct next node in the list correctly.
However, it happened that the fix is way more involved than just fixing
that small issue, due to the way that semaphore group timeouts work.
Instead of handling timeouts one-by-one, we have to handle all timeouts
in a semaphore group as one. To do that, we use the fact that the
timeout of the real thread is always found first in the kernel's
timeout_q, and if it has expired, we do not even look at the timeouts of
the dummy threads.
Change-Id: Iadcfd06f33c6b335efa2592b2c01eeb5ca67afde
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Timeouts, when expiring on the same tick, should be handled in the same
order they were queued.
Change-Id: I23a8e971a47ca056b32b8b48fe179d481bae27c0
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Queuing in the timeout_q of timeouts expiring on the same tick queue
them in reverse order: as soon as the new timeout finds a timeout
expiring on the same tick or later, it get prepended to that timeout:
this allows exiting the traversal of the timeout as soon as possible,
which is done with interrupts locked, thus reducing interrupt latency.
However, this has the side-effect of handling the timeouts expiring on
the same tick in the reverse order that they are queued.
For example:
thread_c, prio 4:
uint32_t uptime = k_uptime_get_32();
while(uptime == k_uptime_get_32()); /* align on tick */
k_timer_start(&timer_a, 5, 0);
k_timer_start(&timer_b, 5, 0);
thread_a, prio 5:
k_timer_status_sync(&timer_a);
printk("thread_a got timer_a\n");
thread_b, prio 5:
k_timer_status_sync(&timer_b);
printk("thread_b got timer_b\n");
One could "reasonably" expect thread_a to run first, since both threads
have the same prio, and timer_a was started before timer_b, thus
inserted first in the timeout_q first (time-wise). However, thread_b
will run before thread_a, since timer_b's timeout is prepended to
timer_a's.
This patch keeps the reversing of the order when adding timeouts in the
timeout_q, thus preserving the same interrupt latency; however, when
dequeuing them and adding them to the expired queue, we now reverse that
order _again_, causing the timeouts to be handled in the expected order.
Change-Id: Id83045f63e2be88809d6089b8ae62034e4e3facb
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Modify _get_first_thread_to_unpend() so that it does not remove the
thread from the wait queue. Rename it to _find_first_thread_to_unpend()
to match the new behaviour.
This will be needed to fix a semaphore group bug.
Change-Id: I1b7531c3beecf3b6a86ecf88a93a02449edd0767
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Rather than explicitely checking the thread state bit.
Change-Id: Ic78427d9847e627a0e91d0147d3b6164450597f6
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
To be API-equivalent with doubly-linked lists.
Change-Id: I98b781f4c649e248abb04f660f686ad76d6b39de
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Like SYS_DLIST_FOR_EACH_NODE(), but __dn contains a node where to fetch
the next node from, NULL to start at the head.
Note that the function does not iterate from @a node, but from
node->next. This allows the following:
sys_dnode_t *funcA(sys_dlist_t *list, sys_dnode_t *node)
{
SYS_DLIST_ITERATE_FROM_NODE(list, node) {
if (node == <some condition>) {
return node;
}
}
return NULL;
}
sys_dlist_t list = &<some list>;
sys_dnode_t *node = NULL;
do {
node = funcA(list, node)
if (node == <some other condition>) {
goto found;
}
} while(node);
<handle error>
found:
<do stuff with node>
Change-Id: I17a5787594a0ed1a4745bd2e1557dd54895105ca
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
This has not bitten us yet, but it was a ticking timebomb.
This is similar to the issue that was found with irq_lock/irq_unlock
implementations on several architectures. Having a volatile variable is
not the way to force the sched_lock variable to be
incremented/decremented around the accesses to data it protects.
Instead, a compiler barrier must prevent the compiler from reordering
the memory accesses around setting of sched_lock. Needed in the inline
implementations _sched_lock()/_sched_unlock_no_reschedule(), which
resolve to simple decrement/increment of the per-thread sched_lock
variable.
Change-Id: I06f5b3524889f193efe69caa947118404b1be0b5
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
This starts development on master for Zephyr 1.8
Change-Id: I81484f09cd4ed90c276f2901be90132af90c2208
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Add a README.rst file to both echo-server and echo-client
sample applications.
Jira: ZEP-1539
Change-Id: I1ee1d02a3de0295697117f448d8e13f5b161e11b
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Currently, for the following MQTT msg fields:
- client_id
- will_topic
- user_name
- topic
their length is computed inside the routine that receives the MQTT msg.
Although this simplifies development, also imposes one restriction:
data must be null-terminated. Sometimes, data is received from other
sources and not generated by the application, so the null-terminated
constraint may be considered problematic for the user.
This patch removes the assumption that string fields are null-terminated.
Current data structures are already prepared to handle this case, so no
API change is required.
Change-Id: I5a147a5b21e0da49541cbe62baac363c8737cd3e
Signed-off-by: Flavio Santes <flavio.santes@intel.com>
This patch updates the Remaining Length field from uint16_t to
uint32_t. The MQTT std specifies that this field must be
unsigned 4 bytes length.
Change-Id: I319d0745c673faece4bbd4db29b1bafad78ac199
Signed-off-by: Flavio Santes <flavio.santes@intel.com>
Add the QEMU x86 prj file to the MQTT publisher application.
Change-Id: I1ba8c3a3bf818d8f6dee7ea94021a83ba5277e5f
Signed-off-by: Flavio Santes <flavio.santes@intel.com>
Duplicate const specifier, it should be the pointed data that is const.
Jira: ZEP-1723
Change-Id: I194abb0fc9ad564c6d53e4727bd63c8099d4eb2e
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
This README file is no longer needed as we have separate
README files in various sub-directories.
Change-Id: I38d1c4208e373aa7aa23373149ef1b89fbcda094
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
This includes:
* Moving the Nordic nRF5x Segger J-Link page (temporarily placed inside
one of the board folders)
* Moving the nrf52_pca10040 doc
* Moving the nrf51_pca10028 doc
* Moving the nrf52840_pca10056 doc
Change-Id: I051eb51cee8166ae6472eb696ffeb0625a0424c7
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
