This is throwing errors in static analysis, complaining that comparing
that a prior is higher and lower is impossible. That is wrong per my
eyes (I swear I think it might be cueing off the names of the
functions, which invert "higher" and "lower" to match our reversed
priority numbers).
But frankly this was never a very readable macro to begin with.
Refactor to put the bounds into the term, so the static analyzer can
prove it locally, and add a build assertion to catch any errors (there
are none currently) where the low<->high priority range is invalid.
Long term, we should probably remove this macro, it doesn't provide
much value. But removing it in response to a static analysis failure
is... not very responsible as a development practice.
Fixes#14816Fixes#14820
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The pointer arithmetic used didn't account for ARC
supervisor mode stacks, which are allocated at the
end of the stack object. Use the new macro to know
exactly how much space is reserved.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Daniel Leung caught a good one: In the (SMP) case where we were
aborting a thread that was not currently scheduled, we were flagging
the DEAD state on _current and not the thread we were aborting! This
wasn't as fatal as it seems, as the thread that called z_sched_abort()
would effectively go on living (as a zombie?) in a state where it
would always be preempted, but would otherwise remain scheduleable.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Most CPUs have instructions like LOCK, LDREX/STREX, etc which
allows for atomic operations without locking interrupts that
can be invoked from user mode without complication. They typically
use compiler builtin atomic operations, or custom assembly
to implement them.
However, some CPUs may lack these kinds of instructions, such
as Cortex-M0 or some ARC. They use these C-based atomic
operation implementations instead. Unfortunately these require
grabbing a spinlock to ensure proper concurrency with other
threads and ISRs. Hence, they will trigger an exception when
called from user mode.
For these platforms, which support user mode but not atomic
operation instructions, the atomic API has been exposed as
system calls.
Some of the implementations in atomic_c.c which can be instead
expressed in terms of other atomic operations have been removed.
The kernel test of atomic operations now runs in user mode to
prove that this works.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The workaround for nonatomic swap had yet another edge case: it would
save off the _current pointer when pending a thread so that the next
time slice interrupt could test it to see if the swap had actually
happened before assuming that _current could be rescheduled (if it
just pended itself, that's impossible). Then it would clear the
pending_current pointer so future interrupts wouldn't be confused.
BUT: it turns out that qemu, when faced with really rapid timer rates
that exceed its (host-based) timing accuracy, is perfectly willing to
"stack up" timer interrupts such the one goes pending before the
previous one is finished executing. In that case, we can enter the
SECOND timer interrupt, to try timeslicing a SECOND time, STILL before
the PendSV exception has run to actually effect the context switch.
Except this time pending_current has been cleared and we try to
reschedule the pended _current thread incorrectly. In theory real
hardware could do this too, though it would involve absolutely crazy
interrupt latency problems.
Work around this by moving the clear to the thread itself, immediately
after it wakes up from the pend call it retakes a lock and clears
pending_current if it still matches _current. That is not a perfect
fix: there remains a 2-3 instruction race at that moment where we
return from pend and before we can lock interrupts again where a timer
interrupt will see an incorrect pointer. But I hammered at this and
couldn't make qemu do that (i.e. return from a timer interrupt but
flag a new one in just a cycle or two).
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The existing device_set_power_state() API works only in synchronous
mode and this is not desirable for devices(ex: Gyro) which take
longer time (few 100 mSec) to suspend/resume.
To support async mode, a new callback argument is added to the API.
The device drivers can asynchronously suspend/resume and call the
callback function upon completion of the async request.
This commit adds the missing callback parameter to all the drivers
to make it compliant with the new API.
Signed-off-by: Ramakrishna Pallala <ramakrishna.pallala@intel.com>
Corrections in the documentation of arguments in
Z_SYSCALL_MEMORY, Z_SYSCALL_MEMORY_READ, and
Z_SYSCALL_MEMORY_WRITE macros.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
This commit enhances the documentation of z_arch_buffer_validate
describing the cases where the validation is performed
successfully, as well as the cases where the result is
undefined.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
This was always doing a remove/add of the _current thread to the run
queue, which is wrong because in SMP _current isn't in the queue to
remove. But it went undetected until the recent dlist changes.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
In SMP, we are setting the _current pointer while holding the
scheduler spinlock locally, which means that when we try to release it
the validation layer (not the spinlock per se) will scream at us
because the thread that took the lock doesn't match the one releasing
it.
Special case this when validation is enabled.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The k_spin_lock() validation was setting the new owner of the spinlock
BEFORE the actual lock was taken, so it could race against other
processors trying the same thing. Split the modification step out
into a separate function that can be called after we affirmatively
have the lock.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The tracing fixes in commit e87193896a ("subsys: debug: tracing: Fix
thread tracing") were... not a readability win. The point appears to
have been to put a tracing hook immediately before and after the
assignment to the _current pointer. So do that in an abstracted
function and clean up _get_next_switch_handle() (which is a subtle and
important function already polluted with some unavoidable preprocessor
testing!)
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Currently thread abort doesn't work if a thread is currently scheduled
on a different CPU, because we have no way of delivering an interrupt
to the other CPU to force the issue. This patch adds a simple
framework for an architecture to provide such an IPI, implements it
for x86_64, and uses it to implement a spin loop in abort for the case
where a thread is currently scheduled elsewhere.
On SMP architectures (xtensa) where no such IPI is implemented, we
fall back to waiting on an arbitrary interrupt to occur. This "works"
for typical code (and all current tests), but of course it cannot be
guaranteed on such an architecture that k_thread_abort() will return
in finite time (e.g. the other thread on the other CPU might have
taken a spinlock and entered an infinite loop, so it will never
receive an interrupt to terminate itself)!
On non-SMP architectures this patch changes no code paths at all.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Tickless timers mean that k_busy_wait() won't work until after the
timer driver is initialized, which is very early but not as early as
SMP. No need for it, just spin.
Also the original code used a stack variable for the start flag, which
racily presumed that _arch_start_cpu() would comes back synchronously
with the other CPU fired up and running right now. The cleaned up smp
bringup API on x86_64 isn't so perky, so it exposed the bug. The flag
just needs to be static.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
In SMP, _current is not "queued". (The run queue only stores
unscheduled threads because we can't rely on the head of the list
being _current). We weren't updating the cache choice, which would
flag swap_ok, so calling k_thread_abort(_current) (for example, when a
thread exits from its entry function) would try to switch back into
the thread and then run off the end of the function.
Amusingly this was more benign than you'd think. Stumbled on it by
accident.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
When CONFIG_SMP is enabled but CONFIG_MP_NUM_CPUS is 1 (which is a
legal configuration, though a weird one) this static function ends up
being defined but unused, producing a compiler warning.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
There was a detected user error in the code where racing insertions of
k_delayed_work items into different queues would be detected and
flagged as an error (honestly I don't see much value there -- Zephyr
doesn't as a general rule protect against errors like this, and
work_q's are inherently kernel things that don't require
userspace-style checking).
This got broken with spinlockification, where each work_q object got
its own lock, so the single lock wouldn't protect against the other
insert function any more. As it happens, that was needless. The core
synchronization on a work_q is in the internal k_queue object anyway
-- the lock in this file was only ever used for (very fast,
noncontending) delayed work insertion. So go back to a global lock to
preserve the original behavior.
Fixes#14104
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Update reserved function names starting with one underscore, replacing
them as follows:
'_k_' with 'z_'
'_K_' with 'Z_'
'_handler_' with 'z_handl_'
'_Cstart' with 'z_cstart'
'_Swap' with 'z_swap'
This renaming is done on both global and those static function names
in kernel/include and include/. Other static function names in kernel/
are renamed by removing the leading underscore. Other function names
not starting with any prefix listed above are renamed starting with
a 'z_' or 'Z_' prefix.
Function names starting with two or three leading underscores are not
automatcally renamed since these names will collide with the variants
with two or three leading underscores.
Various generator scripts have also been updated as well as perf,
linker and usb files. These are
drivers/serial/uart_handlers.c
include/linker/kobject-text.ld
kernel/include/syscall_handler.h
scripts/gen_kobject_list.py
scripts/gen_syscall_header.py
Signed-off-by: Patrik Flykt <patrik.flykt@intel.com>
Rename scheduler spinlock sched_lock to sched_spinlock as it will
collide with the cleanup of the reserved function name _sched_lock(),
which will also be called sched_lock().
Signed-off-by: Patrik Flykt <patrik.flykt@intel.com>
Add a note in the documentatation of _setup_new_thread()
function stating that the caller is responsible for
providing a size argument that corresponds to the availabe
thread stack area.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
This commit forces architecture-specific implementation for
initializing the are for user mode local thread data. This
has been enforced already for ARC. We now do the same for ARM.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
The mbedtls library has some globals which results in faults
when user mode tries to access them.
Instantiate a memory partition for mbedtls's globals.
The linker will place all globals found by building this
library into this partition.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Some init tasks may use some bss app memory areas and
expect them to be zeroed out. Do this much earlier
in the boot process, before any of the init tasks
run.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
We used to leave byte-long placeholder symbols to ensure
that empty application memory sections did not cause
build errors that were very difficult to understand.
Now we use some relatively portable inline assembly to
generate a symbol, but don't take up any extra space.
The malloc and libc partitions are now only instantiated
if there is some data to put in them.
Fixes: #13923
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
On x86, if a supervisor thread belonging to a memory domain
adds a new partition to that domain, subsequent context switches
to another thread in the same domain, or dropping itself to user
mode, does not have the correct setup in the page tables.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
In some circumstances (e.g., a tickless kernel), k_timer_remaining_get()
would not account for time passed that didn't involve clock interrupts.
This adds a simple fix for that, and adds a test case. In addition, the
return value of k_timer_remaining_get() is clamped at 0 in the case of
overdue timers and the API description is adjusted to reflect this.
Fixes: #13353
Signed-off-by: Charles E. Youse <charles.youse@intel.com>
Retpolines were never completely implemented, even on x86.
Move this particular Kconfig to only concern itself with
the assembly code, and don't default it on ever since we
prefer SSBD instead.
We can restore the common kernel-wide CONFIG_RETPOLINE once
we have an end-to-end implementation.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Nonatomic swap strikes again. These issues are all longstanding, but
were unmasked by the dlist work in commit d40b8ce1fb ("sys: dlist:
Add sys_dnode_is_linked") where list node pointers become nulls on
removal.
The previous fix was for a specific case where a timeslicing interrupt
would try to slice out the "wrong" current thread because the thread
has "just" pended itself. That was incomplete, because the parallel
code in k_sleep() didn't flag itself the same way.
And beyond that, it turns out to be basically impossible (now that I'm
thinking about it correctly) to prevent interrupt code from calling
into the scheduler to suspend a "just pended but not quite" current
and/or preempt away to another thread. In any of these cases, the
scheduler modifications to the state bits remain correct but the queue
nodes may be corrupt because the thread was already removed from the
ready queue. So we have to test and correct this at the lowest level,
where a thread is being removed from a priq: check that it's (1) the
ready queue and not a waitq, (2) the current thread, and (3) already
marked suspended and thus not in the queue.
There are lots of existing issues filed in the last few months all
pointing to odd instability on ARM platforms. I'm reasonably certain
this is the root cause for most or all of them.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
CPU_LPS_n name used to indicate a low power state is cryptic and
incorrect. The low power states act on the whole SoC and not exclusively
on the CPU. This patch renames CPU_LPS_n states to LOW_POWER_n. Also
HAS_ pattern for Kconfig options is used in favor of a non standard
_SUPPORTED. Naming of deep sleep states was adjusted accordingly.
Following is a detailed list of string replacements used:
s/SYS_POWER_STATE_CPU_LPS_(\d)_SUPPORTED/HAS_STATE_LOW_POWER_$1/
s/SYS_POWER_STATE_CPU_LPS_(\d)/SYS_POWER_STATE_LOW_POWER_$1/
s/SYS_POWER_STATE_DEEP_SLEEP_(\d)_SUPPORTED/HAS_STATE_DEEP_SLEEP_$1/
Signed-off-by: Piotr Mienkowski <piotr.mienkowski@gmail.com>
This commit removes dependency on SYS_POWER_LOW_POWER_STATES_SUPPORTED,
SYS_POWER_DEEP_SLEEP_STATES_SUPPORTED Kconfig options. Power management
SYS_POWER_LOW_POWER_STATES, SYS_POWER_DEEP_SLEEP_STATES options depend
now directly on specific power states supported by the given SoC. This
simplifies maintenance of SoC Kconfig files.
Signed-off-by: Piotr Mienkowski <piotr.mienkowski@gmail.com>
We can just implement this as a macro and not needlessly
run afoul of MISRC-C rule 17.1
Fixes: #10012
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This is an integral part of userspace and cannot be used
on its own. Fold into the main userspace configuration.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
User mode needs to be able to read this value in
compiler generated function prologues/epilogues.
Special handling in init.c for arches that use
_data_copy. This happens before _Cstart() gets
called. We need to make sure that the compiler
stack canary checks in _data_copy itself do not
fail.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
We need a generic name for the partition containing
essential C library globals. We're going to need to
add the stack canary guard to this area so user mode
can read it.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Instead of having to enable ramfunc support manually, just make it
transparently available to users, keeping the MPU region disabled if not
used to not waste a MPU region. This however wastes 24 bytes of code
area when the MPU is disabled and 48 bytes when it is enabled, and
probably a dozen of CPU cycles during boot. I believe it is something
acceptable.
Note that when XIP is used, code is already in RAM, so the __ramfunc
keyword does nothing, but does not generate an error.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The linker file defines the __ramfunc_ram_size symbols to get the size
of the __ramfunc_ram section. Use that instead of computing the value at
runtime from the start and end symbols. This saves 16 bytes of code with
CONFIG_RAM_FUNCTION=y.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Using __ramfunc to places a function in RAM instead of Flash.
Code that for example reprograms flash at runtime can't execute
from flash, in that case must placing code into RAM.
This commit create a new section named '.ramfunc' in link scripts,
all functions has __ramfunc keyword saved in thats sections and
will load from flash to sram after the system booted.
Fixes: #10253
Signed-off-by: qianfan Zhao <qianfanguijin@163.com>
This commit simplifies OS <-> Application interface controlling power
management. In the previous approach application-based PM required
overriding sys_suspend() and sys_resume() functions. As these functions
actually implemented power state change, in such case application
basically had to provide own implementation of all PM-related stuff,
which was not portable and hard to maintain.
This commit changes this scheme: The sys_suspend() and sys_resume()
are now system functions while the application could either use
built-in power management policies or provide its own. All details
of power mode switching are now handled by the OS.
Also, this commit cleans up the Kconfig options related to system-level
power management grouping them under common CONFIG_SYS_PM_ prefix.
Signed-off-by: Piotr Zięcik <piotr.ziecik@nordicsemi.no>
There are issues using lowercase min and max macros when compiling a C++
application with a third-party toolchain such as GNU ARM Embedded when
using some STL headers i.e. <chrono>.
This is because there are actual C++ functions called min and max
defined in some of the STL headers and these macros interfere with them.
By changing the macros to UPPERCASE, which is consistent with almost all
other pre-processor macros this naming conflict is avoided.
All files that use these macros have been updated.
Signed-off-by: Carlos Stuart <carlosstuart1970@gmail.com>
Work queues are implemented in terms of k_queue objects which provide
their own synchronization. In particular insertion is potentially
blocking and always acts as a reschedule point, which means that it
must not be called with spinlocks held.
Release the lock first, and do a little cleanup of the resulting
k_delayed_work_submit_to_queue() logic.
Fixes#13411
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Upon hard/soft irq or exception entry/exit, handle transitions
off or onto the trampoline stack, which is the only stack that
can be used on the kernel side when the shadow page table
is active. We swap page tables when on this stack.
Adjustments to page tables are now as follows:
- Any adjustments for stack memory access now are always done
to the user page tables
- Any adjustments for memory domains are now always done to
the user page tables
- With KPTI, resetting a page now clears the present bit
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Minor style (syntax) fix in the help text of symbol
config EXECUTION_BENCHMARKING.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
This commit changes the names of SYS_POWER_DEEP_SLEEP* Kconfig
options in order to match SYS_POWER_LOW_POWER_STATE* naming
scheme.
Signed-off-by: Piotr Zięcik <piotr.ziecik@nordicsemi.no>
