This replaces the existing scheduler (but not priority handling)
implementation with a somewhat simpler one. Behavior as to thread
selection does not change. New features:
+ Unifies SMP and uniprocessing selection code (with the sole
exception of the "cache" trick not being possible in SMP).
+ The old static multi-queue implementation is gone and has been
replaced with a build-time choice of either a "dumb" list
implementation (faster and significantly smaller for apps with only
a few threads) or a balanced tree queue which scales well to
arbitrary numbers of threads and priority levels. This is
controlled via the CONFIG_SCHED_DUMB kconfig variable.
+ The balanced tree implementation is usable symmetrically for the
wait_q abstraction, fixing a scalability glitch Zephyr had when many
threads were waiting on a single object. This can be selected via
CONFIG_WAITQ_FAST.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This was wrong in two ways, one subtle and one awful.
The subtle problem was that the IRQ lock isn't actually globally
recursive, it gets reset when you context switch (i.e. a _Swap()
implicitly releases and reacquires it). So the recursive count I was
keeping needs to be per-thread or else we risk deadlock any time we
swap away from a thread holding the lock.
And because part of my brain apparently knew this, there was an
"optimization" in the code that tested the current count vs. zero
outside the lock, on the argument that if it was non-zero we must
already hold the lock. Which would be true of a per-thread counter,
but NOT a global one: the other CPU may be holding that lock, and this
test will tell you *you* do. The upshot is that a recursive
irq_lock() would almost always SUCCEED INCORRECTLY when there was lock
contention. That this didn't break more things is amazing to me.
The rework is actually simpler than the original, thankfully. Though
there are some further subtleties:
* The lock state implied by irq_lock() allows the lock to be
implicitly released on context switch (i.e. you can _Swap() with the
lock held at a recursion level higher than 1, which needs to allow
other processes to run). So return paths into threads from _Swap()
and interrupt/exception exit need to check and restore the global
lock state, spinning as needed.
* The idle loop design specifies a k_cpu_idle() function that is on
common architectures expected to enable interrupts (for obvious
reasons), but there is no place to put non-arch code to wire it into
the global lock accounting. So on SMP, even CPU0 needs to use the
"dumb" spinning idle loop.
Finally this patch contains a simple bugfix too, found by inspection:
the interrupt return code used when CONFIG_SWITCH is enabled wasn't
correctly setting the active flag on the threads, opening up the
potential for a race that might result in a thread being scheduled on
two CPUs simultaneously.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The smp_init() call was too early. Device and subsystem
initialization doesn't happen until after the main thread starts
running. Starting extra CPUs and allowing them to schedule threads
before their drivers are alive is a bad idea, even if it works in a
unit test.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This uses the version and hash (git describe) and replaces the timestamp
currently used in the boot banner. This works much better than using
timestamps. It lets us point to the exact commit being used to run a
certain application or test.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The scheduler exposed two APIs to do the same thing:
_add_thread_to_ready_q() was a low level primitive that in most cases
was wrapped by _ready_thread(), which also (1) checks that the thread
_is_ready() or exits, (2) flags the thread as "started" to handle the
case of a thread running for the first time out of a waitq timeout,
and (3) signals a logger event.
As it turns out, all existing usage was already checking case #1.
Case #2 can be better handled in the timeout resume path instead of on
every call. And case #3 was probably wrong to have been skipping
anyway (there were paths that could make a thread runnable without
logging).
Now _add_thread_to_ready_q() is an internal scheduler API, as it
probably always should have been.
This also moves some asserts from the inline _ready_thread() wrapper
to the underlying true function for code size reasons, otherwise the
extra use of the inline added by this patch blows past code size
limits on Quark D2000.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Names that begin with an underscore are reserved by the C standard.
This patch does not change names of functions defined and implemented
in header files.
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
Now that all the pieces are in place, enable SMP for real:
Initialize the CPU records, launch the CPUs at the end of kernel
initialization, have them wait for a flag to release them into the
scheduler, then enter into the runnable threads via _Swap().
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The scheduler needs a few tweaks to work in SMP mode:
1. The "cache" field just doesn't work. With more than one CPU,
caching the highest priority thread isn't useful as you may need N
of them at any given time before another thread is returned to the
scheduler. You could recalculate it at every change, but that
provides no performance benefit. Remove.
2. The "bitmask" designed to prevent the need to individually check
priorities is likewise dropped. This could work, but in fact on
our only current SMP system and with current K_NUM_PRIOPRITIES
values it provides no real benefit.
3. The individual threads now have a "current cpu" and "active" flag
so that the choice of the next thread to run can correctly skip
threads that are active on other CPUs.
The upshot is that a decent amount of code gets #if'd out, and the new
SMP implementations for _get_highest_ready_prio() and
_get_next_ready_thread() are simpler and smaller, at the expense of
having to drop older optimizations.
Note that scheduler synchronization is unchanged: all scheduler APIs
used to require that an irq_lock() be held, which means that they now
require the global spinlock via the same API. This should be a very
early candidate for lock granularity attention!
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Simple implementation that caps at 4 CPUs. Long term we should use
some linker magic to define as many as needed and loop over them
without needlessly increasing data or code size for the tracking.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The xtensa-asm2 work included a patch that added nano_internal.h
includes in lots of places that needed to have _Swap defined, because
it had to break a cycle and this no longer got pulled in from the arch
headers.
Unfortunately those new includes created new and more amusing cycles
elsewhere which led to breakage on other platforms.
Break out the _Swap definition (only) into a separate header and use
that instead. Cleaner. Seems not to have any more hidden gotchas.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
_Swap() is defined in nano_internal.h. Everything calls _Swap().
Pretty much nothing that called _Swap() included nano_internal.h,
expecting it to be picked up automatically through other headers (as
it happened, from the kernel arch-specific include file). A new
_Swap() is going to need some other symbols in the inline definition,
so I needed to break that cycle. Now nothing sees _Swap() defined
anymore. Put nano_internal.h everywhere it's needed.
Our kernel includes remain a big awful yucky mess. This makes things
more correct but no less ugly. Needs cleanup.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
For the dummy thread, contents in the mem_domain structure
is insignificant hence setting it to NULL.
Signed-off-by: Adithya Baglody <adithya.nagaraj.baglody@intel.com>
Kernel object metadata had an extra data field added recently to
store bounds for stack objects. Use this data field to assign
IDs to thread objects at build time. This has numerous advantages:
* Threads can be granted permissions on kernel objects before the
thread is initialized. Previously, it was necessary to call
k_thread_create() with a K_FOREVER delay, assign permissions, then
start the thread. Permissions are still completely cleared when
a thread exits.
* No need for runtime logic to manage thread IDs
* Build error if CONFIG_MAX_THREAD_BYTES is set too low
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Some "random" drivers are not drivers at all: they just implement the
function `sys_rand32_get()`. Move those to a random subsystem in
preparation for a reorganization.
Signed-off-by: Leandro Pereira <leandro.pereira@intel.com>
Intention of CONFIG_BOOT_DELAY is to delay booting of system for certain
time. Currently it is only delaying start of _main thread as delay is
created using k_sleep. This leads to putting _main thread into timeout
queue and continue kernel boot. This is causing some of undesirable
effects in some of test Automation usecase.
This patch changes k_sleep to k_busy_wait which result in delay in OS
boot instead of delaying start of _main.
Signed-off-by: Youvedeep Singh <youvedeep.singh@intel.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
It's currently too easy to run out of thread IDs as they
are never re-used on thread exit.
Now the kernel maintains a bitfield of in-use thread IDs,
updated on thread creation and termination. When a thread
exits, the permission bitfield for all kernel objects is
updated to revoke access for that retired thread ID, so that
a new thread re-using that ID will not gain access to objects
that it should not have.
Because of these runtime updates, setting the permission
bitmap for an object to all ones for a "public" object doesn't
work properly any more; a flag is now set for this instead.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Now creating a thread will assign it a unique, monotonically increasing
id which is used to reference the permission bitfield in the kernel
object metadata.
Stub functions in userspace.c now implemented.
_new_thread is now wrapped in a common function with pre- and post-
architecture thread initialization tasks.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Garbage values here could wreak havoc on the initial switch to main
depending on how arch-specific _Swap() manages memory permissions when
switching threads.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
All system calls made from userspace which involve pointers to kernel
objects (including device drivers) will need to have those pointers
validated; userspace should never be able to crash the kernel by passing
it garbage.
The actual validation with _k_object_validate() will be in the system
call receiver code, which doesn't exist yet.
- CONFIG_USERSPACE introduced. We are somewhat far away from having an
end-to-end implementation, but at least need a Kconfig symbol to
guard the incoming code with. Formal documentation doesn't exist yet
either, but will appear later down the road once the implementation is
mostly finalized.
- In the memory region for RAM, the data section has been moved last,
past bss and noinit. This ensures that inserting generated tables
with addresses of kernel objects does not change the addresses of
those objects (which would make the table invalid)
- The DWARF debug information in the generated ELF binary is parsed to
fetch the locations of all kernel objects and pass this to gperf to
create a perfect hash table of their memory addresses.
- The generated gperf code doesn't know that we are exclusively working
with memory addresses and uses memory inefficently. A post-processing
script process_gperf.py adjusts the generated code before it is
compiled to work with pointer values directly and not strings
containing them.
- _k_object_init() calls inserted into the init functions for the set of
kernel object types we are going to support so far
Issue: ZEP-2187
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Fixes https://github.com/zephyrproject-rtos/zephyr/issues/1280, but
also many other failures, where output was garbled due to this. Other
similarly affected issues are missing first benchmark (context) in
latency benchmark and some net tests.
Signed-off-by: Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com>
The API/Variable names in timing_info looks very speicific to
platform (like systick etc), whereas these variabled are used
across platforms (nrf/arm/quark).
So this patch :-
1. changing API/Variable names to generic one.
2. Creating some of Macros whose implimentation is platform
depenent.
Jira: ZEP-2314
Signed-off-by: Youvedeep Singh <youvedeep.singh@intel.com>
The boot banner is being printed after static threads have started, for
example this is visible with tests using ztest.
This puts the banner message before starting any threads.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
* add nested interrupt support for interrupts
+ use a varibale exc_nest_count to trace nest interrupt and exception
+ regular interrupts can be nested by regular interrupts and fast
interrupts
+ fast interrupt's priority is the highest, cannot be nested
* remove the firq stack and exception stack
+ remove the coressponding kconfig option
+ all interrupts (normal and fast) and exceptions will be handled
in the same stack (_interrupt stack)
+ the pros are, smaller memory footprint (no firq stack), simpler
stack management, simpler codes, etc.. The cons are, possible
10-15 instructions overhead for the case where fast irq nests
regular irq
* add the case of ARC in test/kernel/gen_isr_table
Signed-off-by: Wayne Ren <wei.ren@synopsys.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Subsequent patches will set this guard page as unmapped,
triggering a page fault on access. If this is due to
stack overflow, a double fault will be triggered,
which we are now capable of handling with a switch to
a know good stack.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Introduce a configurable boot delay option (defaulting to none) that
happens right after printing a boot delay banner, #before calling
main() in kernel/init.c:_main(), before taking timestamps for _main()
and once all the infrastructure is in place. Move also the boot banner
to happen after this delay.
The rationale for this is some boards will boot really fast and print
out some test case output in the serial port before the system that is
monitoring the serial port is able to read from the serial port.
This happens in MCUs whose serial port is embedded in a USB connection
which also is used to power the MCU board. When powering it on by
powering the USB port, there is a time it takes the host system to
detect the USB connection, enumerate the serial port, configure it and
load, start and read from the serial port. At this time, it might have
printed the output of the serial port.
While manually it is possible to press a reset button, on automation
setups this adds a lot of overhead and cabling or modifications to the
MCU that are easier (and cheaper) to overcome with this delay. Other
options (like using a separate serial line) might not be possible or
add a lot of cabling and cost, plus it'd also add extra build
configuration.
Change-Id: I2f4d1ba356de6cefa19b4ef5c9f19f87885d4dfd
Signed-off-by: Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com>
Applications will have their own BSS and data sections which
will need to be additionally copied.
This covers the common C implementation of these functions.
Arches which implement their own optimized versions will need
to be updated.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
1. Changed _tsc_read() to k_cycles_get_32(). Thus reading the
time stamp will be agnostic of the architecutre used.
2. Changed the variable names from *_tsc to *_time_stamp.
JIRA: ZEP-1426
Signed-off-by: Adithya Baglody <adithya.nagaraj.baglody@intel.com>
The existing __stack decorator is not flexible enough for upcoming
thread stack memory protection scenarios. Wrap the entire thing in
a declaration macro abstraction instead, which can be implemented
on a per-arch or per-SOC basis.
Issue: ZEP-2185
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The initial dummy thread context used for the initial __swap to
the main thread at early kernel initialization was not marked as a dummy
thread as it ought to be.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Unline k_thread_spawn(), the struct k_thread can live anywhere and not
in the thread's stack region. This will be useful for memory protection
scenarios where private kernel structures for a thread are not
accessible by that thread, or we want to allow the thread to use all the
stack space we gave it.
This requires a change to the internal _new_thread() API as we need to
provide a separate pointer for the k_thread.
By default, we still create internal threads with the k_thread in stack
memory. Forthcoming patches will change this, but we first need to make
it easier to define k_thread memory of variable size depending on
whether we need to store coprocessor state or not.
Change-Id: I533bbcf317833ba67a771b356b6bbc6596bf60f5
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Convert code to use u{8,16,32,64}_t and s{8,16,32,64}_t instead of C99
integer types. This handles the remaining includes and kernel, plus
touching up various points that we skipped because of include
dependancies. We also convert the PRI printf formatters in the arch
code over to normal formatters.
Jira: ZEP-2051
Change-Id: Iecbb12601a3ee4ea936fd7ddea37788a645b08b0
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Filter was wrong and sample was not being built on any boards. Exclude
platforms that do not support interrupt based UART drivers.
Jira: ZEP-2014
Change-Id: I84a690e7c93fae52335434830b83086019cfd00d
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Due to a limitation on XCC, the inline assembly does not
produce the expected instructions. This results in a wrong
code sequence. On the other hand, plain C code works well.
The note about compilers seems to not be an issue on any of
our currently supported compilers.
Change-Id: I9d2ab0fbf8a48d9dad51da3fd54453f205516d74
Signed-off-by: Mazen NEIFER <mazen@nestwave.com>
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
The K_<thread option> flags/options avaialble to users were hidden in
the kernel private header files: move them to include/kernel.h to
publicize them.
Also, to avoid any future confusion, rename the k_thread.execution_flags
field to user_options.
Change-Id: I65a6fd5e9e78d4ccf783f3304b607a1e6956aeac
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
The execution_flags will store the user-facing states of a thread.
This also fixes a bug where K_ESSENTIAL was already assigned to
execution_flags via the options field of
k_thread_spawn()/K_THREAD_DEFINE().
Change-Id: I91ad7a62b5d180e09eead8985ff519809959ecf2
Signed-off-by: Benjamin Walsh <walsh.benj@gmail.com>
Replace the existing Apache 2.0 boilerplate header with an SPDX tag
throughout the zephyr code tree. This patch was generated via a
script run over the master branch.
Also updated doc/porting/application.rst that had a dependency on
line numbers in a literal include.
Manually updated subsys/logging/sys_log.c that had a malformed
header in the original file. Also cleanup several cases that already
had a SPDX tag and we either got a duplicate or missed updating.
Jira: ZEP-1457
Change-Id: I6131a1d4ee0e58f5b938300c2d2fc77d2e69572c
Signed-off-by: David B. Kinder <david.b.kinder@intel.com>
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
added _MOVE_INSTR for RISCV32 architecture
The store instruction has a different syntax in RISC-V,
compared to the other architectures. Hence, for each
architecture, specify the entire load instruction within
the _MOVE_INSTR variable.
Change-Id: Iedc421e73411876abd8b698f7d4b46081b473d79
Signed-off-by: Jean-Paul Etienne <fractalclone@gmail.com>
The main, idle, interrupt and workqueue call stack definitions are not available
to applications to call stack_analyze() on, but they often require to be
measured empirically to tune their sizes in particular applications and
use cases.
This exposes a new k_call_stacks_analyze() API call that allows the
application to measure the used call stack space for the 4
kernel-defined call stacks.
Additionally for the ARC architecture the FIRQ stack is also profiled.
Change-id: I0cde149c7366cb6c4bbe8f9b0ab1cc5b56a36ed9
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Some thread fields were 32-bit wide, when they are not even close to
using that full range of values. They are instead changed to 8-bit fields.
- prio can fit in one byte, limiting the priorities range to -128 to 127
- recursive scheduler locking can be limited to 255; a rollover results
most probably from a logic error
- flags are split into execution flags and thread states; 8 bits is
enough for each of them currently, with at worst two states and four
flags to spare (on x86, on other archs, there are six flags to spare)
Doing this saves 8 bytes per stack. It also sets up an incoming
enhancement when checking if the current thread is preemptible on
interrupt exit.
Change-Id: Ieb5321a5b99f99173b0605dd4a193c3bc7ddabf4
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
Also remove mentions of unified kernel in various places in the kernel,
samples and documentation.
Change-Id: Ice43bc73badbe7e14bae40fd6f2a302f6528a77d
Signed-off-by: Anas Nashif <anas.nashif@intel.com>