On 64-bit targets, the minimum possible mempool block size is not 8
but 16. With a max block size of 32, the mempool allocator cannot
split it into 4 sub-blocks, reducing the available memory allocations
to that original 32-byte block only.
To get the same allocation patterns and test behavior whether it is
built for a 32-bit or 64-bit architecture, let's define BLK_SIZE_MIN
and BLK_SIZE_MAX in terms of _MPOOL_MINBLK instead of literal values.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Documentation for k_pipe_block_put() says:
This routine writes the data contained in a memory block to pipe.
Once all of the data in the block has been written to the pipe,
it will free the memory block.
Therefore it is wrong to free the memory block within the test code.
When the mempool allocator is instrumented to detect double-free
instances, this case is signaled right away.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Re-run with updated script to convert integer literal delay arguments
to k_thread_create and K_THREAD_DEFINE to use the standard timeout
macros.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Re-run with updated script to convert integer literal delay arguments to
k_sleep to use the standard timeout macros.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Re-run with updated script to convert integer literal delay arguments to
k_mbox_data_block_get to use the standard timeout macros.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Just housekeeping around the casting between void * arguments to
thread functions and integer types.
Signed-off-by: Charles E. Youse <charles.youse@intel.com>
This was a very early test and got bitrotten inside a esp32-only
whitelist. Make it run generically.
SMP must be forced off by the test (it's commonly a platform default).
Add a build-time failure when the configuration is single-CPU, for
clarity.
Filter the test likewise so it runs on all supported systems.
Also, the key argument to the CPU startup function is vestigial and
the test was being too strict by requiring it to be non-zero.
Finally, the qemu command line needs to predicate the "-smp" argument
on CONFIG_MP_NUM_CPUS and not just CONFIG_SMP so we have an extra CPU
to test against.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This test was written very early. Spinlocks are required for SMP
implementation. They couldn't be tested in terms of it, so the test
used the low level MP API instead. But of course that breaks if SMP
is actually working and the CPU is already started.
No need for that now. Just spawn a thread like any other, and filter
the test to run only on SMP systems.
Fixes#19319
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This patch is a preparatory step in enabling the MMU in
long mode; no steps are taken to implement long mode support.
We introduce struct x86_page_tables, which represents the
top-level data structure for page tables:
- For 32-bit, this will contain a four-entry page directory
pointer table (PDPT)
- For 64-bit, this will (eventually) contain a page map level 4
table (PML4)
In either case, this pointer value is what gets programmed into
CR3 to activate a set of page tables. There are extra bits in
CR3 to set for long mode, we'll get around to that later.
This abstraction will allow us to use the same APIs that work
with page tables in either mode, rather than hard-coding that
the top level data structure is a PDPT.
z_x86_mmu_validate() has been re-written to make it easier to
add another level of paging for long mode, to support 2MB
PDPT entries, and correctly validate regions which span PDPTE
entries.
Some MMU-related APIs moved out of 32-bit x86's arch.h into
mmustructs.h.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This commit adds new k_work_poll interface. It allows to
submit given work to a workqueue automatically when one of the
watched pollable objects changes its state.
Signed-off-by: Piotr Zięcik <piotr.ziecik@nordicsemi.no>
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
Use the int_literal_to_timeout Coccinelle script to convert literal
integer arguments for kernel API timeout parameters to the standard
timeout value representations.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
This test requires more than 32 static priorities by default, and
doesn't run with the multiq scheduler without a special configuration.
That used to be specified per-platform, but got moved to a separate
test case a while back.
This broke non-default platforms like qemu_cortex_m3 which use
SCHED_MULTIQ as their default backend. Put a filter in place instead
of going back to per-platform changes.
Fixes#19437
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Remove duplicate "tdata.timestamp" update in duration_expire; this
value is already updated by k_uptime_delta.
Besides simply removing duplicate value update, this commit also
addresses the intermittent assertion failure that is caused by
updating "tdata.timestamp" at a later time than the actual execution
of the k_uptime_delta function.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
We filter out the following kernel tests
- tickless_concept
- timer_api
from the set of tests running on QEMU Cortex-M0 platform,
as the tests consistently fail on QEMU. In addition, we
add a workaround for kernel/interrupt test, so it can
successfully execute on QEMU Cortex-M0.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
Increased TX/RX buffer size by one in pipe test to prevent buffer
overrun.
Some test will transfer one byte more then the number of bytes
supported by the pipe, in case the buffer size is the same as the
size of the pipe this will result in a buffer overrun.
Tools such as address sanitizer would detect this overrun and fail the
test.
Signed-off-by: Jan Van Winkel <jan.van_winkel@dxplore.eu>
The main and idle threads, and their associated stacks,
were being referenced in various parts of the kernel
with no central definition. Expose these in kernel_internal.h
and namespace with z_ appropriately.
The main and idle threads were being defined statically,
with another variable exposed to contain their pointer
value. This wastes a bit of memory and isn't accessible
to user threads anyway, just expose the actual thread
objects.
Redundance MAIN_STACK_SIZE and IDLE_STACK_SIZE defines
in init.c removed, just use the Kconfigs they derive
from.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This is part of the core kernel -> architecture interface
and is appropriately renamed z_arch_is_in_isr().
References from test cases changed to k_is_in_isr().
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Disabling SMP mode for certain tests was a one-release thing, done to
avoid having to triage every test independently (MANY are not
SMP-safe), and with the knowledge that it was probably hiding bugs in
the kernel.
Turn it on pervasively. Tests are treated with a combination of
flagging specific cases as "1cpu" where we have short-running tests
that can be independently run in an otherwise SMP environment, and via
setting CONFIG_MP_NUM_CPUS=1 where that's not possible (which still
runs the full SMP kernel config, but with only one CPU available).
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
This test was testing for an undocumented and somewhat hyperspecific
behavior: when a process reaches a reschedule point and yields to a
higher priority thread, and there is another equal priority thread
active, which thread gets to run when the higher priority thread
finishes its work? The original scheduler (because it leaves the
older thread in place in the list) implements the preemption like an
interrupt and returns to the original thread, despite the fact that
this then resets is time slice quantum unfairly. In SMP mode, where
the current threads cannot live in the active list, the thread gets
added back to the end of the queue and the other thread runs. In
effect, in UP mode "yield" and "reschedule" mean very slightly
different things where in SMP they act the same.
We don't document either behavior, as it happens. Relax the test
constraints by adding a single deliberate k_yield() to unify behavior.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
To catch more potential issues with PRs, build common kernel tests
in addition to the synchronization sample which does not run any tests.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Platforms with limited flash are now failing to link. Add or increase
flash requirements for test cases to exclude the ones that will fail.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
The RTC on TI CC13X2/CC26X2 is a 32 KHz clock for which the minimum
compare delay is 3 ticks. When using it as the system clock, we need
to relax the upper bound to ensure the test succeeds.
Signed-off-by: Vincent Wan <vincent.wan@linaro.org>
In stack_sentinel_timer(), the timer should not be allocated on the
stack. If it gets added to the list of timeouts by k_timer_start,
then an unexpected exception may occur when the timer expires since it
may have been overwritten.
Signed-off-by: Vincent Wan <vincent.wan@linaro.org>
This driver was still using CONFIG_* values to determine its address,
IRQ, etc. Add a binding for an "intel,hpet" device and migrate this
driver to devicetree.
Fixes: #18657
Signed-off-by: Charles E. Youse <charles.youse@intel.com>
Suppress integer overflow warning generated by the check macros
NEG_CHECK and ROLLOVER_CHECK in intmath tests
Signed-off-by: Jan Van Winkel <jan.van_winkel@dxplore.eu>
These two tests were right at the knife edge of 16kb
on riscv64, and were not building with logging enabled
on that platform.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This test for some reason wants to validate that
k_stack_analyze() works when called from the idle thread,
but with a default idle stack size of 256 this just results
in crashes.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
System call arguments, at the arch layer, are single words. So
passing wider values requires splitting them into two registers at
call time. This gets even more complicated for values (e.g
k_timeout_t) that may have different sizes depending on configuration.
This patch adds a feature to gen_syscalls.py to detect functions with
wide arguments and automatically generates code to split/unsplit them.
Unfortunately the current scheme of Z_SYSCALL_DECLARE_* macros won't
work with functions like this, because for N arguments (our current
maximum N is 10) there are 2^N possible configurations of argument
widths. So this generates the complete functions for each handler and
wrapper, effectively doing in python what was originally done in the
preprocessor.
Another complexity is that traditional the z_hdlr_*() function for a
system call has taken the raw list of word arguments, which does not
work when some of those arguments must be 64 bit types. So instead of
using a single Z_SYSCALL_HANDLER macro, this splits the job of
z_hdlr_*() into two steps: An automatically-generated unmarshalling
function, z_mrsh_*(), which then calls a user-supplied verification
function z_vrfy_*(). The verification function is typesafe, and is a
simple C function with exactly the same argument and return signature
as the syscall impl function. It is also not responsible for
validating the pointers to the extra parameter array or a wide return
value, that code gets automatically generated.
This commit includes new vrfy/msrh handling for all syscalls invoked
during CI runs. Future commits will port the less testable code.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
XIP support in x86 was something of a mess. This
patch does the following:
- Generic ia32 SOC no longer defines a "flash" region
as generic X86 devices don't have a microcontroller-
like concept of flash. The same has been done for apollo_lake.
- Generic ia32 and apollo_lake SOCs starts memory at 1MB.
- Generic ia32 SOC may optionally have CONFIG_XIP enabled.
The board definition must provide a flash region definition
that gets exposed as DT_PHYS_LOAD_ADDR.
- Fixed definitions for RAM/ROM source addresses in ia32's
linker.ld when XIP is turned off.
- Support for enabling XIP on apollo_lake SOC removed, there's
no use-case.
- acrn and gpmrb boards have flash and XIP related definitions
removed.
- qemu_x86 has a fake flash region added, immediately after system
RAM, for use when XIP is enabled. This used to be in the ia32 SOC.
However, the default for qemu_x86 is to now have XIP disabled.
- Fixed tests/kernel/xip to run by default on boards that enable
XIP by default, plus an additional test to exercise XIP on
qemu_x86 (which supports it but has XIP switched off by default)
The overall effect of this patch is to:
- Remove XIP configuration for SOC/boards where it does not make
any sense to have it
- Support testing XIP on qemu_x86 via tests/kernel/xip, but leave
it off by default for other tests, to ensure it doesn't bit-rot
and that the system works in both scenarios.
- XIP remains an available feature for boards that need it.
Fixes: #18956
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Several user mode tests cannot run on twr_ke18f because
either the platform does not have a sufficient number of
MPU regions required for the tests, or, the tests also
require HW stack protection (which has been, by default,
excluded in user mode tests for twr_ke18f board). We
excluded the board from all those tests.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
This commit replaces several CONFIG_USERSPACE=y
settings with CONFIG_TEST_USERSPACE=y. This allows
the test sub-system Kconfig structure to control
the settings of USERSPACE and HW_STACK_PROTECTION
in the various tests suites.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
As we have re-worked the test code, and the test-case can run
on Cortex-M platforms on any available and implemented NVIC
IRQ lines, we do not need to exclude these ARM boards anymore.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
This commit re-works the test for the ARM architecture,
so that it can work with any available NVIC IRQ, not
bound to use the last 2 NVIC lines. It makes use of
the dynamic IRQ feature.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
Fix unhandled return values as most other places handled in this
file, fix coverity issue 203507.
Fixes: #18445.
Signed-off-by: Wentong Wu <wentong.wu@intel.com>
Fix unhandled return values as most other places handled in this
file, fix coverity issue 203454.
Fixes: #18443.
Signed-off-by: Wentong Wu <wentong.wu@intel.com>
Nordic platforms measure ticks in cycles of a 32 KiHz clock for which
the minimum compare delay is 2 ticks. The test assumed an upper bound
of four ticks delay per loop iteration, resulting from alignment at
various layers. This delay is met on Nordic for tick rates at or below
16384, but is too short for the default 32768 Hz tick rate.
Instrumentation confirms that the usleep test loop body on Nordic at 32
KiHz ticks takes 3 ticks as the optimum delay, only when a debug probe
is active. In other circumstances it can take either 5 or 6 ticks,
depending on timer alignment and stability.
Relax the upper bound for platforms using this system timer at the
highest rate.
Closes#17965.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
