In the timeout order test, the usage of k_poll() assumes that it
only returns after all events are ready. However, that is not
the case, as k_poll() returns when non-zero number of events are
ready. This means the check for all semaphore being ready after
k_poll() will not always pass. So instead of using k_poll(),
simply wait a bit for timers to fire, then check results.
Also add some bits to clean up at the end of test.
Fixes#34585
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
It's not at all clear to me why this was set to 1cpu, it's a single
thread doing sequential things. (I tripped over it because the 1cpu
happened to tickle an unrelated arm64 bug with interrupt state. But
we might as well fix it here.)
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Kernel objects that contain embedded synchronization structures like
spinlocks can't be palced in the (cached/incoherent) stack memory on
coherence platforms like intel_adsp.
The normal fix in a test case is just to make the offending data
static, but that's painful here because SYS_BITARRAY_DEFINE declares
two objects (i.e. you can't put a "static" in front of it as with
similar macros) and it happens to be used in this case to define local
variables with collliding names, so I'd have to go in and rename
everything.
And there's little value anyway. Bitarrays are nearly-pure data
structures and extremely unlikely to show up platform-dependent
behavior.
Fixes#35242
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Treat ARMV7_M_ARMV8_M_MAINLINE similarly to ARMV6_M_ARMV8_M_BASELINE
and add arch_nop() calls to test_nop function.
Additionally add one arch_nop() call to fit comment and update
comments when required on other archs.
Signed-off-by: Erwan Gouriou <erwan.gouriou@linaro.org>
The newly added testcase test_nop failed the CI. Give RISCV more
arch_nop() instructions to archieve one cycle.
Signed-off-by: Enjia Mai <enjiax.mai@intel.com>
new test failed which means we missed something in CI or the failing
platform changed after CI was initially run. skip it for now while we
investigate.
Do some minor cleanup in the metadata.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Add a test case to test arch interface arch_nop(), the main focus here
is for coverage of the code. arch_nop() is a special implementation
and it will behave differently on different platforms. By the way, this
also measures how many cycles it spends for platforms that support it.
Signed-off-by: Enjia Mai <enjiax.mai@intel.com>
Add an testcase. Creat two preempt threads with equal priority to
atomiclly access the same atomic value. Because these preempt
threads are of equal priority, so enable time slice to make
them scheduled. The thread will execute for some time.
In this time, the two sub threads will be scheduled separately
according to the time slice.
Signed-off-by: Ying ming <mingx.ying@intel.com>
First, this test is a little suspect. It's assuming that the value
returned from k_cycle_get_32() represents the time since system
power-on. While that's an obvious implementation choice and surely
often true, it's definitely not the way we document this API to the
arch layer. It's perfectly legal for a platform to return any value
as long as the counter is increasing at the correct rate. Leaving for
now as there's no other way to test CONFIG_BOOT_DELAY, but this will
likely be coming back to confuse us at some point.
Regardless, that convention holds for x86 devices using any of the
existing drivers. But on an EFI PC using the TSC counter as the clock
source: (1) the counter is running at 1-2 GHz and (2) the time to get
through an EFI BIOS and into Zephyr is routinely 10+ seconds,
especially on reference hardware. The poor 32 bit API will roll over
several times, and effectively be a random number by the time it
reaches this test.
Just skip this test with fast counter.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Add QEMU board with single core ARCv3 HS6x 64 bit CPU
Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
Signed-off-by: Evgeniy Paltsev <PaltsevEvgeniy@gmail.com>
This adds some tests to make sure sys_bitarray_*() are
working correctly.
Signed-off-by: Lauren Murphy <lauren.murphy@intel.com>
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
There's a typedef for non-pointer values compatible with atomic
non-pointer objects. Add a similar typedef for pointer values, and
the corresponding macro for initializing atomic pointer types.
This also will simplify replacing the Zephyr atomic API with one
based on C11 atomics, should that be desirable. C11 atomic pointer
values are not void*.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
This functions is being called across the tree, no reason why it should
not be a public API.
The current usage violates a few MISRA rules.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
A fairly common idiom in our test code is to put test-local data
structures onto the stack, even when they are to be used from another
thread. But stacks are incoherent memory on some platforms, which
means that such things may not get a consistent view of memory between
threads.
Just make these things static. A few of these spots were causing test
failures on intel_adsp_cavs15. More were found by inspection while
hunting for mistakes.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
tls rely on both arch has tls and toolchain support tls, add filter:
CONFIG_TOOLCHAIN_SUPPORTS_THREAD_LOCAL_STORAGE for
some tests enabled tls.
Signed-off-by: Watson Zeng <zhiwei@synopsys.com>
build_on_all here was supposed to be a smoke test to test building on
all platforms, it should not be used for more than 1 just test.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The BIT_INDEX() macro assumed little-endian. This commit adds
big-endian support, conditioned on the preprocessor define
CONFIG_BIG_ENDIAN.
Signed-off-by: Martin Åberg <martin.aberg@gaisler.com>
Using the same implementation as the rest of Zephyr reduces code size.
Update options and expected results for formatting test.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
These tests were suppressed when KERNEL_COHERENCE=y because of a
feature collision with CONFIG_POLL that has since been fixed.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The existing testcase's doxygen describes are the general
implementation idea of a function.On this basis, adding
more descriptive statements to describe which conditions need
to be preset when running the testcase, which test techniques
are applied, and describe the testcase Design steps in detail.
Make it more readable.
Signed-off-by: Ying ming <mingx.ying@intel.com>
The k_poll implementation places a struct _poller on the stack and
shares it with other threads, which is incompatible with the
KERNEL_COHERENCE model of cached stacks.
Make this a hard build failure instead of a kconfig dependency for
clarity. The failures if a user actually enables both are subtle and
difficult to debug.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Reduce the error between the timer (which is tick-aligned) and
busy_wait (which is not) by aligning the busy_wait to start at
a tick boundary.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
The width for %p on 32-bit targets should be 8 regardless of
CONFIG_PRINTK64. Adjust the test accordingly.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
After the Qemu Cortex-M0 timer driver rework, we may
re-enable the test-cases that had been skipped for this
platform.
Signed-off-by: Ioannis Glaropoulos <Ioannis.Glaropoulos@nordicsemi.no>
Add support for 64 bit conversions in a uniformly expressable way by
printing values backwards into a buffer on the stack first. This
allows all operations to work on the low bits of the value and so the
code doesn't need to care (beyond the size of that buffer) about the
word size. This trick also doesn't care about the specifics of the
base value, so in the process this unifies the decimal and hex printk
conversion code to a single function.
This comes at a mild cost in CPU cycles to the decimal converter and
somewhat higher cost to hex (because it's now doing a full div/mod
operation instead of shifting and masking). And stack usage has grown
by a few words to hold the temporary. But the benefits in code size
are substantial (e.g. ~250 bytes of .text on arm32).
Note that this also contains a change to tests/kernel/common to
address what appears to have been a bug in the original converters.
The printk test uses a format string that looks like "%-4x%-2p" and
feeds it the literal arguments "0xABCDEF" and "(char *)42".
Now... clearly both those results are going to overflow the 4 and
2-byte field sizes, so there shouldn't be any whitespace between these
fields. But the test was written to expect two spaces, inexplicably
(yes, I checked: POSIX-compatible printf implementations don't have
those spaces either).
The new code is definitely doing the right thing, so fix the test
instead.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
add new test cases to illustrate the zephyr OS
support an array of atomic variables, each bit
of which can be modified.
Signed-off-by: Ying ming <mingx.ying@intel.com>
Test fails on this one platform, to unblock other changes, skip the test
while the issues is being looked at. This test never ran on this
platform before due to ram restrictions.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
rename boot_delay function name for clarity and change doxygen group to
be more generic and part of the init group.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Remove all ram restrictions in kernel tests and revisit all tests and
try to make them pass on all platforms.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
... because it is (required).
This makes a difference when building with CMake and forgetting
ZEPHYR_BASE or not registering Zephyr in the CMake package registry.
In this particular case, REQUIRED turns this harmless looking log
statement:
-- Could NOT find Zephyr (missing: Zephyr_DIR)
-- The C compiler identification is GNU 9.3.0
-- The CXX compiler identification is GNU 9.3.0
-- Check for working C compiler: /usr/bin/cc
-- ...
-- ...
-- ...
-- Detecting CXX compile features
-- Detecting CXX compile features - done
CMake Error at CMakeLists.txt:8 (target_sources):
Cannot specify sources for target "app" which is not built by
this project.
... into this louder, clearer, faster and (last but not least) final
error:
CMake Error at CMakeLists.txt:5 (find_package):
Could not find a package configuration file provided by "Zephyr" with
any of the following names:
ZephyrConfig.cmake
zephyr-config.cmake
Add the installation prefix of "Zephyr" to CMAKE_PREFIX_PATH or set
"Zephyr_DIR" to a directory containing one of the above files. If
"Zephyr" provides a separate development package or SDK, be sure it
has been installed.
-- Configuring incomplete, errors occurred!
Signed-off-by: Marc Herbert <marc.herbert@intel.com>
This test works by starting a bunch of poll events, dropping the test
thread priority, calling k_poll(), and assuming that all the timeouts
that fired woke up high priority threads and thus ran before k_poll()
could return. But that isn't true if you have another CPU that can
run the low priority thread while the last high priority thread
finishes up!
This just isn't SMP-correct. Mark 1cpu.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Kernel timeouts have always been a 32 bit integer despite the
existence of generation macros, and existing code has been
inconsistent about using them. Upcoming commits are going to make the
timeout arguments opaque, so fix things up to be rigorously correct.
Changes include:
+ Adding a K_TIMEOUT_EQ() macro for code that needs to compare timeout
values for equality (e.g. with K_FOREVER or K_NO_WAIT).
+ Adding a k_msleep() synonym for k_sleep() which can continue to take
integral arguments as k_sleep() moves away to timeout arguments.
+ Pervasively using the K_MSEC(), K_SECONDS(), et. al. macros to
generate timeout arguments.
+ Removing the usage of K_NO_WAIT as the final argument to
K_THREAD_DEFINE(). This is just a count of milliseconds and we need
to use a zero.
This patch include no logic changes and should not affect generated
code at all.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Using find_package to locate Zephyr.
Old behavior was to use $ENV{ZEPHYR_BASE} for inclusion of boiler plate
code.
Whenever an automatic run of CMake happend by the build system / IDE
then it was required that ZEPHYR_BASE was defined.
Using ZEPHYR_BASE only to locate the Zephyr package allows CMake to
cache the base variable and thus allowing subsequent invocation even
if ZEPHYR_BASE is not set in the environment.
It also removes the risk of strange build results if a user switchs
between different Zephyr based project folders and forgetting to reset
ZEPHYR_BASE before running ninja / make.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
The documentation motivates this function by saying it is more
efficient than the core 64-bit version. This was untrue when
originally added, and is untrue now. Mark the function deprecated and
replace its sole in-tree use with the trivial equivalent.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Remove leading/trailing blank lines in .c, .h, .py, .rst, .yml, and
.yaml files.
Will avoid failures with the new CI test in
https://github.com/zephyrproject-rtos/ci-tools/pull/112, though it only
checks changed files.
Move the 'target-notes' target in boards/xtensa/odroid_go/doc/index.rst
to get rid of the trailing blank line there. It was probably misplaced.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Mark the old time conversion APIs deprecated, leave compatibility
macros in place, and replace all usage with the new API.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Promote the private z_arch_* namespace, which specifies
the interface between the core kernel and the
architecture code, to a new top-level namespace named
arch_*.
This allows our documentation generation to create
online documentation for this set of interfaces,
and this set of interfaces is worth treating in a
more formal way anyway.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.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>
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 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>
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>
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>
An inverted comparison typo led to the final loop in the sflist being
skipped. Fix so that it actually runs.
(Odd that it took a static analysis tool to detect this, the loop
expressions are all constants, I'm surprised gcc didn't see it while
doing unrolling analysis).
Fixes#18437
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The test for the k_uptime_delta utilities was calling it in a loop and
waiting for the uptime to advance. But the code was specifically
wanting it to advance 5ms or more at one go, which clearly isn't going
to work for a tick rate above 200 Hz.
The weird thing is that the test knew this and even commented about
the limitation. Which seems silly: it's perfectly fine for the clock
to advance just a single millisecond. That's correct behavior too.
Let's test for that, and it will work everywhere.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
sflists have a couple APIs related to sfnodes that aren't
present for slists. There were uncovered, write some tests
for them.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
We were testing all the slist APIs, but not the sflist
variant. Make a copy of the slist tests for sflist,
with the names properly changed.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Some of the slist APIs were only being indirectly exercised;
add to the slist test case to cover everything explicitly.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Needed an explicit test for this function for code
coverage purposes; we were relying indirectly on
other code using it.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
move misc/slist.h to sys/slist.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move misc/printk.h to sys/printk.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move misc/dlist.h to sys/dlist.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move misc/byteorder.h to sys/byteorder.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move atomic.h to sys/atomic.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
We had plenty of coverage for k_cycle_get(), but not its
32-bit variant. Run a case in user mode so that the system
call handler gets covered.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
On 64-bit systems the most notable difference is due to longs and
pointers being 64-bit wide. Therefore there must be a distinction
between ints and longs. Similar to the prf.c case, this patch properly
implements the h, hh, l, ll and z length modifiers as well as some small
cleanups.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Using void pointers as universal arguments is widely used. However, when
compiling a 64-bit target, the compiler doesn't like when an int is
converted to a pointer and vice versa despite the presence of a cast.
This is due to a width mismatch between ints (32 bits) and pointers
(64 bits). The trick is to cast to a widening integer type such as
intptr_t and then cast to
void*.
When appropriate, the INT_TO_POINTER macro is used instead of this
double cast to make things clearer. The converse with POINTER_TO_INT
is also done which also serves as good code annotations.
While at it, remove unneeded casts to specific pointer types from void*
in the vicinity, and move to typed variable upon function entry to make
the code cleaner.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Per guidelines, all statements should have braces around them. We do not
have a CI check for this, so a few went in unnoticed.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
It's useful to be able to inspect the key returned from
z_arch_irq_unlock() to see if interrupts were enabled at the point
where z_arch_irq_lock() was called. Architectures tend to represent
this is a simple way that doesn't require platform assembly to
inspect.
Adds a simple test to kernel/common that validates this predicate with
a nested lock.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Update the files which contain no license information with the
'Apache-2.0' SPDX license identifier. Many source files in the tree are
missing licensing information, which makes it harder for compliance
tools to determine the correct license.
By default all files without license information are under the default
license of Zephyr, which is Apache version 2.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Unlike CONFIG_HW_STACK_PROTECTION, which greatly helps
expose stack overflows in test code, activating
userspace without putting threads in user mode is of
very limited value.
Now CONFIG_TEST_USERSPACE is off by default. Any test
which puts threads in user mode will need to set
CONFIG_TEST_USERSPACE.
This should greatly increase sanitycheck build times
as there is non-trivial build time overhead to
enabling this feature. This also allows some tests
which failed the build on RAM-constrained platforms
to compile properly.
tests/drivers/build_all is a special case; it doesn't
put threads in user mode, but we want to ensure all
the syscall handlers compile properly.
Fixes: #15103 (and probably others)
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This lets us quickly filter tests that exercise userspace
when developing it.
Some tests had a whitelist with qemu_cortex_m3; change
this to mps2_an385, which is the QEMU target with an
MPU enabled.
Signed-off-by: Andrew Boie <andrew.p.boie@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>
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>
Fix multiple definitions of `ram_console'. The ram_console
array is already defined in drivers/console/ram_console.c.
Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@st.com>
(Chunk 1 of 3 - this patch was split across pull requests to address
CI build time limitations)
Zephyr has always been a uniprocessor system, and its kernel tests are
rife with assumptions and outright dependence on single-CPU operation
(for example: "low priority threads will never run until this high
priority thread blocks" -- not true if there's another processor to
run it!)
About 1/3 of our tests fail right now on x86_64 when dual processor
operation is made default. Most of those can probably be recovered on
a case-by-case basis with simple changes (and a few of them might
represent real bugs in SMP!), but for now let's make sure the full
test suite passes by turning the second CPU off. There's still plenty
of SMP coverage in the remaining cases.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
MISRA rules (see #9892) forbid alloca() and family, even though those
features can be valuable performance and memory size optimizations
useful to Zephyr.
Introduce a MISRA_SANE kconfig, which when true enables a gcc error
condition whenever a variable length array is used.
When enabled, the mempool code will use a theoretical-maximum array
size on the stack instead of one tailored to the current pool
configuration.
The rbtree code will do similarly, but because the theoretical maximum
is quite a bit larger (236 bytes on 32 bit platforms) the array is
placed into struct rbtree instead so it can live in static data (and
also so I don't have to go and retune all the test stack sizes!).
Current code only uses at most two of these (one in the scheduler when
SCHED_SCALABLE is selected, and one for dynamic kernel objects when
USERSPACE and DYNAMIC_OBJECTS are set).
This tunable is false by default, but is selected in a single test (a
subcase of tests/kernel/common) for coverage. Note that the I2C and
SPI subsystems contain uncorrected VLAs, so a few platforms need to be
blacklisted with a filter.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
The sys_dlist_insert_*() functions had a behavior where a NULL
argument for the insertion position to sys_dlist_insert_after/before()
was interpreted as "the end of the list". We never used that
convention (except in one spot internal to dlist.h which was not
itself used anywhere), and of course already have an API for appending
and prepending to a list.
In practice this was a performance disaster. The NULL check is
virtually never provable statically by the compiler, so that test and
branch is present always. And worse, the check and call to another
function was pushing this beyond the complexity limit for gcc to
inline a function (at -Os optimization anyway), forcing us to use
function calls for what should be a ~8 instruction sequence. The
upshot is that dlist insertions were 2-3x slower than they needed to
be.
Deprecate these older APIs and introduce a new sys_dlist_insert() call
which can be much better optimized.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Several places in the code have constructions like this:
if (bool_variable) {
atomic_set_bit(flags, FLAG);
} else {
atomic_clear_bit(flags, FLAG);
}
To reduce the amount of code for such situations, introduce a new
atomic_set_bit_to() helper which lets you condense the above five
lines to a single one:
atomic_set_bit_to(flags, FLAG, bool_variable);
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
The original implementation allows a list to be corrupted by list
operations on the removed node. Existing code attempts to avoid this by
using external state to determine whether a node is in a list, but this
is fragile and fails when the state that holds the flag value is changed
after the node is removed, e.g. in preparation for re-using the node.
Follow Linux in invalidating the link pointers in a removed node. Add
API so that detection of particpation in a list is available at the node
abstraction.
This solution relies on the following steady-state invariants:
* A node (as opposed to a list) will never be adjacent to itself in a
list;
* The next and prev pointers of a node are always either both null or
both non-null.
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
