Rename _MsrRead() and _MsrWrite() to _x86_msr_read() and
_x86_msr_write() respectively.
Given that these functions are essentially implemented in assembly.
make them static inline. They can be inlined by the compiler quite
well, most of the time incurring in space savings due to better
handling of the cobbled registers.
Also simplifies the inline assembly, using constraints instead of
moving registers ourselves. Should shave off a few bytes from code
using these functions.
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
Very simple implementation of deadline scheduling. Works by storing a
single word in each thread containing a deadline, setting it (as a
delta from "now") via a single new API call, and using it as extra
input to the existing thread priority comparison function when
priorities are equal.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This patch adds a set of priorities at the (numerically) lowest end of
the range which have "meta-irq" behavior. Runnable threads at these
priorities will always be scheduled before threads at lower
priorities, EVEN IF those threads are otherwise cooperative and/or
have taken a scheduler lock.
Making such a thread runnable in any way thus has the effect of
"interrupting" the current task and running the meta-irq thread
synchronously, like an exception or system call. The intent is to use
these priorities to implement "interrupt bottom half" or "tasklet"
behavior, allowing driver subsystems to return from interrupt context
but be guaranteed that user code will not be executed (on the current
CPU) until the remaining work is finished.
As this breaks the "promise" of non-preemptibility granted by the
current API for cooperative threads, this tool probably shouldn't be
used from application code.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The scheduler rewrite added a regression in uniprocessor mode where
cooperative threads would be unexpectedly preempted, because nothing
was checking the preemption status of _current at the point where the
next-thread cache pointer was being updated.
Note that update_cache() needs a little more context: spots like
k_yield() that leave _current runable need to be able to tell it that
"yes, preemption is OK here even though the thread is cooperative'.
So it has a "preempt_ok" argument now.
Interestingly this didn't get caught because we don't test that. We
have lots and lots of tests of the converse cases (i.e. making sure
that threads get preempted when we expect them to), but nothing that
explicitly tries to jump in front of a cooperative thread.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This symbol started out as CONFIG_BLUETOOTH_MAX_CMD_LEN, which was
removed in commit 50678b03cb ("Bluetooth: Reuse HCI command buffers
for the command response").
The non-existing symbol in the assignment was then renamed to
CONFIG_BT_MAX_CMD_LEN by commit 2975ca0754 ("Bluetooth: Kconfig:
Rename CONFIG_BLUETOOTH_* to CONFIG_BT_*").
Remove the assignment.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
The prj.conf file was using wrong VLAN tag option names.
The correct names are CONFIG_SAMPLE_VLAN_TAG and
CONFIG_SAMPLE_VLAN_TAG_2 as defined in Kconfig file.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Remove unused context switch code from the helper assembly file that
manages the SuperVisor Call invocations (SVCs). This is no longer in use
in the code, since all calls to __swap() now go through PendSV directly.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
When we introduced NEWLIB_LIBC_ALIGNED_HEAP_SIZE in commit
42a2c96422. We accidently had the Kconfig
symbol depend on CONFIG_MPU_REQUIRES_POWER_OF_TWO_ALIGNMENT the leading
'CONFIG_' shouldn't exist.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
CONFIG_COMMAND_STACK_SIZE was removed in 4f798177cf ("kernel: remove
old micro/nanokernel C code"). Its help was "Microkernel server command
stack size (in packets)".
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
We had a typo in the Kconfig symbol that was being used to try and set
SYS_LOG_LEVEL. It should be CONFIG_SYS_LOG_NET_LOOPBACK_LEVEL.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
These were using the older name CONFIG_NET_NBUF_DATA_SIZE (note the
extra 'N'). The symbol was renamed in commit bf964cdd4c ("net:
Renaming net nbuf API to net pkt API").
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
The SPI_DW_INTERRUPT_SEPARATED_LINES symbol was removed and replaced by
SPI_DW_PORT_{0,1,2,3}_INTERRUPT_SINGLE_LINE in commit 423f0095c7
("drivers/spi: Specify options per-port on DW driver"), but an
assignment to it remains.
Remove the assignment and set SPI_DW_PORT_0_INTERRUPT_SINGLE_LINE to 'n'
instead. The Arduino has a single SPI port.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
This symbol does not exist. Not sure what was intended.
There's HAS_SEGER_RTT, but that symbol has no prompt, meaning .config
values have no effect on it. It is 'select'ed by the SoCs.
Remove the assignment, which is a no-op.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Not only removes a branch during normal operation, but also ensures
that, by initializing at the PRE_KERNEL_2 stage, and granting
privileges to all threads to the semaphore, this code will work in
early boot situations and in user mode.
This assumes that entropy drivers will all initialize during
PRE_KERNEL_1 stage. All in-tree drivers do that.
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
To ensure that early code that requires entropy has the HWRNG devices
fully initialized, initialize them all during PRE_KERNEL_1 stage.
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
The initial user stack pointer gets rounded to STACK_ALIGN_SIZE,
which can be much finer granularity than MPU regions. With
certain stack size values passed to k_thread_create(), the stack
pointer can be set past the defined region for the user thread's
stack, causing an immediate MPU fault.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Upon return from a syscall handlers, the r1, r2, and r3 registers
could contain random kernel data that should not be leaked to user
mode. Zero these out before returning from _arm_do_syscall().
Fixes#7753.
The invocation macros need a clobber if r1, r2, or r3 are not used
to carry syscall arguments. This is a partial fix for #7754 but
there appear to be other issues.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Many boards do not enable user mode in their default configuration
even though they support it. However for purposes of our unit
testing we really do want it on all the time.
It's not enough to just flip on CONFIG_USERSPACE, the tests largely
assume that CONFIG_APPLICATION_MEMORY and CONFIG_DYNAMIC_OBJECTS
are enabled. Update CONFIG_TEST_USERSPACE to be on by default if
CONFIG_TEST is active, and hang CONFIG_DYNAMIC_OBJECTS on it too.
This problem was not previously detected as qemu_x86 has the
user mode options in its defconfig and it is currently the only
user mode compatible platform which runs in emulation.
Fixes#7692
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This is a workaround until #7704 is fixed, we are getting
an assembler error due to the r7 clobber in the syscall
invocation macros.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
An errant commit accidentally disabled all testing of
hardware-based stack protection. Restore it, and work
around a problem with how these kinds of exceptions are
reported on ARM until #7706 is fixed.
We need to globally disable user mode due to how the
select statements in Kconfig work, the stack sentinel
is incompatible with user mode.
Some build warnings when compiling as native_posix
fixed.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
prepare_multithreading() was done very early as it had a call
to initialize the interrupt subsystem. This was causing problems
with stack pointer randomization as any HW-based entropy drivers
had not been initialized.
Move the call to initialize the interrupt system out of
prepare_multithreading(), which now really does just prepare
the system to start threads. This is now done after the PRE_KERNEL
phases.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
We run into the limit of 32 object files on ARM when
CONFIG_APPLICATION_MEMORY is enabled.
Bug #7703 filed, meanwhile just disable it, it's not
needed for this test case.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
On some arches like ARC, the member location tag is a list with
the offset and then the member size. We just need the offset.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
A test was trying to add the maximum number of partitions,
but when the domain was initialized there was already one
added which needed to be accounted for to avoid an
assertion failing.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The change changes the type of the attribute count in the
bt_gatt_service struct to size_t.
The background for the change is that with this variable being a
u16_t, we assign something that has been cast to an unsigned long
(from "ARRAY_SIZE") to an u16_t variable, and that Lint complains
about a loss of information (27 bits to 16 bits in the case I found).
(The issue seems to be not only about the cast, but about what is
casted. I suspect that Lint may be confused by the magic of
ZERO_OR_COMPILE_ERROR.)
Signed-off-by: Asbjørn Sæbø <asbjorn.sabo@nordicsemi.no>
Turn 'content', 'kconfig', and 'prep' into phony targets so that the
docs can be built even if there is a file called e.g. 'kconfig' in
zephyr/doc/.
Also remove the Python scripts as prerequisites from the rules and
reference them directly in the recipes instead. The phony targets are
always "out-of-date", and there are no rules to run for the Python
scripts themselves.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Make all menu paths ("(top menu) -> menu -> submenu -> ...") exclude
implicit submenus, which are shown indented in the menuconfig interface
and are created when items depend on the symbol before them.
Previously, implicit submenus were excluded in the menu path at the top
of the menuconfig interface, but were included in the menuconfig symbol
information dialog and in the docs generated by genrest.py.
This makes it consistent, and un-spams the documentation for some
symbols a bit.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
i.MX7 updated doc sections for Zephyr release 1.12:
Architectures:
nxp_imx/mcimx7_m4: Added support for i.MX7 Cortex M4 core
Boards:
colibri_imx7d_m4: Added support for Toradex Colibri i.MX7 board
Drivers and Sensors:
serial: Added support for i.MX UART interface
gpio: Added support for i.MX GPIO
HALs:
nxp/imx: imported i.MX7 FreeRTOS HAL
Signed-off-by: Diego Sueiro <diego.sueiro@gmail.com>
Move *.rst to end of the list to ensure all .rst files get reviewed by
our tech writer (last match takes precedence).
Signed-off-by: David B. Kinder <david.b.kinder@intel.com>
The Kconfig option TOOLCHAIN_VARIANT (not to be confused with
ZEPHYR_TOOLCHAIN_VARIANT) is a legacy configuration option that has
very few use-cases and can easily be dropped.
It's functionality is easily covered by CONFIG_X86_IAMCU and
ZEPHYR_TOOLCHAIN_VARIANT.
This commit removes all references of it from Zephyr.
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
Ninja and GNU make don't play well with each other. Both try to start
enough processes to keep the system's CPUs busy, resulting in an
O(N^2) system load in the number of processors.
Long term, Ninja seems likely to support the GNU make jobserver
mechanism for sharing access to parallelism. But for now we can get
90% of the way there with a simple hack: just run ninja in serial mode
with -j1. Sanitycheck when run in non-trivial circumstances has
PLENTY of parallelism just from the number of test cases.
One interesting note is that even with -j1, system loads under ninja
are rather higher. That may be because of significant work done in
the (serial) makefiles that dilutes the parallelism of the eventual
build, or possibly because ninja itself is multithreaded in its setup
code. So I tweaked the number of jobs down to keep the load roughly
where it is with make.
With this change, I see no difference in behavior or system load, and a
~24% improvement in runtime.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
QMSI bindings were created prior to this base, and unfortunately not
updated to latest changes on last rebase.
Fixes#7694
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
The code was using MSEC() macro in few places instead of more
proper K_MSEC(). The MSEC() takes seconds as a parameter and
K_MSEC() takes milliseconds.
Fixes#7657
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Probably left over from old ZOAP library. Stripping of headers
not required now.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
