Make it possible to "finish" with fewer bytes than what was "claimed".
This was possible before on the get side, but the put side was
cummulative wrt finish. The revamp made it cummulative on both sides.
Turns out that existing users rely on the opposite behavior which is
more logical and useful. So make both sides that way.
Adjust documentation, test case and users accordingly.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
The item mode is really a specialization of the byte mode. And in-tree
usage shows the byte mode is prominent. It feels more natural if the
byte mode is presented first with the item mode second. Swap the code
and documentation order accordingly. No code change.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
The core ring buffer code presumes a 32-bit wrap-around behavior when its
head and tail indices cross the maximum value range. Make sure this is
covered by the test suite, for both 32- and 64-bit targets.
While at it, reduce stack requirement of the modified test.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
This code is rather hairy. When I look at it I don't like the way it
stares back at me.
First, the rewind business looks fishy. It has to die.
And we don't have to rely on modulus either. Not even for non-power-of-2
buffers. Let's kill that distinction too and make all sizes always
"high performance".
The code is now entirely relying only on simple ALU operations (add,
sub and compare).
The key assumption: 32-bit values do wrap around after max range has
been reached. No saturation. All architectures supported by Zephyr
do that.
Some stats:
lib/os/ring_buffer.c: 62 insertions(+), 124 deletions(-)
ring_buffer.c.obj before after diff
----------------------------------------------
frdm_k64f 1224 1136 -88
m2gl025_miv 2485 2079 -406
mps2_an385 1228 1132 -96
mps2_an521 1228 1132 -96
native_posix 1546 1496 -50
native_posix_64 1598 1595 -3
nsim_hs_mpuv6 1252 1192 -60
nsim_hs_smp 1252 1192 -60
nsim_sem 1252 1192 -60
qemu_arc_em 1324 1192 -132
qemu_arc_hs6x 1824 1620 -204
qemu_arc_hs 1252 1192 -60
qemu_cortex_a53_smp 2154 1888 -266
qemu_cortex_a53 2154 1888 -266
qemu_cortex_a9 1938 1792 -146
Before (qemu_cortex_a53):
START - test_ringbuffer_performance
1 byte put-get, avg cycles: 52
4 byte put-get, avg cycles: 47
1 byte put claim-finish, avg cycles: 39
5 byte put claim-finish, avg cycles: 41
5 byte get claim-finish, avg cycles: 52
PASS - test_ringbuffer_performance in 0.8 seconds
After (qemu_cortex_a53):
START - test_ringbuffer_performance
1 byte put-get, avg cycles: 34
4 byte put-get, avg cycles: 41
1 byte put claim-finish, avg cycles: 27
5 byte put claim-finish, avg cycles: 29
5 byte get claim-finish, avg cycles: 29
PASS - test_ringbuffer_performance in 0.4 seconds
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Conceptually, ring_buf_item_put() and ring_buf_item_get() are specialized
versions of ring_buf_put() and ring_buf_get(). Make it so to rationalize
the code to open the way for more optimizations.
This means we need specialized wrappers on top of ring_buf_init()
accordingly, given that the core machinery is now common and byte based.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
Clarify that the reader MUST track the last sequence number read and
that the 'start' offset is not involved in that.
Some pre-conceptions + the sentence "the current state of the stream"
threw me off for a while. Thanks Andy for explaining this to me.
Signed-off-by: Marc Herbert <marc.herbert@intel.com>
Add API to fetch current buffer usage. Add option to track
maximum buffer usage and API to fetch that value.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Clang introduced _Generic support in 3.0 but it was buggy until
3.8. Updated guard to not use _Generic in clang < 3.8.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
XCC (which is based on GCC 4.2) needs the initializer of one of the
union elements to be enclosed in brackets.
Signed-off-by: Maureen Helm <maureen.helm@intel.com>
As described in #42403, there was an issue with the existing crc16_ansi()
implementation, since it was not calculating the CRC-16-ANSI (aka
CRC-16-MODBUS). This is because the existing crc16() function only
supported non-reflected input and output (and the CRC-16-ANSI requires
reflection on both) and also it did not seem to support correctly inial
seeds different from 0x0000 (and, again, the CRC-16-ANSI requires 0xffff
as an initial seed).
This commit replaces the existing crc16() with a functional pair,
crc16() and crc16_reflect(), that also work with any poly, any initial seed
and allow to select whether reflection is performed.
It also adapts crc16_ansi() so that it actually returns the correct CRC.
Fixes#42403.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
The crc16_ccitt and crc16_itu_t functions' documentation could lead to
slight misunderstandings because of the fact that, although they bear
those names, those functions are actually able to compute CRCs that bear
other names, depending on the inital seed provided and an optional XOR
on the result.
In order to avoid confusion, document thoroughly with links to the
standard reveng page that lists all existing CRCs.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
This adds to the cbprintf_packaged library to allow external
formatters to be used by the way of callback. This will allow
logging backends to use their own formatter for output if
needed.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
In order to use functions orginating from C libs in C++ context, their
definitions must be declared as "C" functions
Signed-off-by: Kim Bøndergaard <kim@fam-boendergaard.dk>
Fixed a dependency from printk.h to logging headers which in
certain configurations could lead to circular dependencies.
Cleaned up printk.c to call z_log_vprintk from vprintk.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Add maximum timeout used for conversion to Kconfig. Option is used
to determine which conversion algorithm to use: faster but overflowing
earlier or slower without early overflow.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
Avoid result overflow due to intermediate product overflow.
Algorithm was multiplying input value by target frequency
before dividing it by source frequency. If target frequency was
high (e.g. conversion to nanoseconds) it could easily lead to
overflow even though final result would not overflow. Adjusting
algorithm to avoid that.
Note, that typically this code is resolved at compile time so
it will not impact performance as long as it can be resolved.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
It's not uncommon to have Zephyr running in environments where it
shares a memory bus with a foreign/non-Zephyr system (both the older
Intel Quark and cAVS audio DSP systems share this property). In those
circumstances, it would be nice to have a utility that allows an
arbitrary-sized chunk of that memory to be used as a unidirectional
buffered byte stream without requiring complicated driver support.
sys_winstream is one such abstraction.
This code is lockless, it makes no synchronization demands of the OS
or hardware beyond memory ordering[1]. It implements a simple
file/socket-style read/write API. It produces small code and is high
performance (e.g. a read or write on Xtensa is about 60 cycles plus
one per byte copied). It's bidirectional, with no internal Zephyr
dependencies (allowing it to be easily ported to the foreign system).
And it's quite a bit simpler (especially for the reader) than the
older cAVS trace protocol it's designed to replace.
[1] Which means that right now it won't work reliably on arm64 until
we add a memory barrier framework to Zephyr! See notes in the code;
the locations for the barriers are present, but there's no utility to
call.
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
For functions returning nothing, there is no need to document
with @return, as Doxgen complains about "documented empty
return type of ...".
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
For functions returning nothing, there is no need to document
with @return, as Doxgen complains about "documented empty
return type of ...".
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This adds similar ability of sys_multi_heap to the memory blocks
allocator, where a choice function can be used to select
which allocator (of a group) is used for memory block allocation.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This introduces yet another memory blocks allocator where:
() All memory blocks have a single fixed size.
() Multiple blocks can be allocated or freed at the same time.
() A group of blocks allocated together may not be contiguous.
This is useful for operations such as scatter-gather DMA
transfers.
() Bookkeeping of allocated blocks is done outside of
the associated buffer (unlike memory slab). This allows
the buffer to reside in memory regions where these can be
powered down to conserve energy.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Add a function to copy a UTF-8 encoded string that
ensure correct truncation of the string if the source
is larger than the destination, as well as ensuring that
the resulting destination string is NULL-terminated.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
Add a function that can properly truncate UTF-8 strings
without leaving unterminated started characters,
as UTF-8 characters can be 1-4 bytes long.
Signed-off-by: Emil Gydesen <emil.gydesen@nordicsemi.no>
The header can't be fully used in standalone mode: toolchain.h has to be
included first, otherwise the ALWAYS_INLINE attribute is not defined.
Headers that can be directly included and are not self-contained should
be considered a bad practice.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
This extends the heap_listener to cover more events,
specifically, allocation, free and realloc.
Note that typedef are used so the callback can be
documented.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Add a parameter to the resize callback to also take the heap ID.
This allows a single callback to be used for multiple heaps if
so desired.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
In include/sys/device_mmio.h, replacing <toolchain/common.h>
fixes the following warning:
$ west build -b qemu_x86 tests/arch/x86/static_idt/
In file included from zephyr/include/toolchain.h:50,
from zephyr/include/linker/section_tags.h:12,
from zephyr/include/linker/sections.h:132,
from zephyr/include/sys/device_mmio.h:19,
from zephyr/include/drivers/interrupt_controller/loapic.h:14,
from zephyr/include/drivers/interrupt_controller/sysapic.h:10,
from zephyr/include/arch/x86/arch.h:231,
from zephyr/include/arch/cpu.h:15,
from zephyr/tests/arch/x86/static_idt/src/test_stubs.S:17:
zephyr/include/toolchain/gcc.h:61: error: BUILD_ASSERT redefined [-Werror]
61 | #define BUILD_ASSERT(EXPR, MSG...) _Static_assert(EXPR, "" MSG)
|
In file included from zephyr/include/sys/device_mmio.h:18,
from zephyr/include/drivers/interrupt_controller/loapic.h:14,
from zephyr/include/drivers/interrupt_controller/sysapic.h:10,
from zephyr/include/arch/x86/arch.h:231,
from zephyr/include/arch/cpu.h:15,
from zephyr/tests/arch/x86/static_idt/src/test_stubs.S:17:
zephyr/include/toolchain/common.h:165: note: this is the location of the
previous definition
165 | #define BUILD_ASSERT(EXPR, MSG...) \
<toolchain.h> provides a compiler-specific BUILD_ASSERT.
<toolchain/common.h> provides a generic, fallback BUILD_ASSERT and
should probably never be included directly.
Thanks to Gerard Marull-Paretas for recommending this fix.
Related to commit af20208cd9 ("devices: mark device MMIO declarations
to boot/pinned sections") that added #include <linker/sections.h>
Signed-off-by: Marc Herbert <marc.herbert@intel.com>
This is meant to be called by drivers on PCIe BUS with a fixed (from
DTS) IRQ or an MSI one. I.e.: all drivers without a runtime IRQ
detection/allocation.
The arch-specific macro is required to be implemented relevantly.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
The docstring for this function states that it is guaranteed to perform
a copy byte by byte, but this is not true in general without a
`volatile` storage type on the casted pointer.
Signed-off-by: Chris McDonald <cjmcdonald@chromium.org>
Current usage was being limited to PCIE but there are new use case
that demand for this feature to be generalized.
It is made arch-specific since allocation policy may vary
per-architecture.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
* add generic heap event listener module that can be used
for notifying an application of heap-related events
* use the listener module in newlib libc hooks
* add a unit test
Signed-off-by: Damian Krolik <damian.krolik@nordicsemi.no>
Several points check the value of CONFIG_s without first checking that
they're defined. This causes an warning when adding -Wundef to the
build.
Signed-off-by: Yuval Peress <peress@google.com>
This adds the architecture interface so that the GDB stub can
deal with breakpoints and watchpoints. By default, weak
functions are implemented to indicate breakpoints and
watchpoints are not supported.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This adds architecture-specific functions to read/write registers.
This allows architecture to have a sparse representation of
the register file as not all registers are saved during context
switches. This saves some runtime space, and provides some
flexibility on what architectures can do.
Remove from header the need to define ARCH_GDB_NUM_REGISTERS as
it is no longer used in the common gdbstub code.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Add flags to macro which checks if string must be packaged
using runtime approach.
Added flag CBPRINTF_MUST_RUNTIME_PACKAGE_CONST_CHAR. When flag
is set then const char pointers are considered as pointers to
fixed strings and do not require runtime packaging.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
This changes the arch_mem_domain_*() functions to return errors.
This allows the callers a chance to recover if needed.
Note that:
() For assertions where it can bail out early without side
effects, these are converted to CHECKIF(). (Usually means
that updating of page tables or translation tables has not
been started yet.)
() Other assertions are retained to signal fatal errors during
development.
() The additional CHECKIF() are structured so that it will bail
early if possible. If errors are encountered inside a loop,
it will still continue with the loop so it works as before
this changes with assertions disabled.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
With this change, `atomic_t` is 32-bit for 32-bit architectures
and 64-bit for 64-bit architectures. More specifically,
`sizeof(atomic_t) == sizeof(long)`.
Fixes#39530
Signed-off-by: Christopher Friedt <chrisfriedt@gmail.com>
add functions to get the sys_heap runtime statistics,
include total free bytes, total allocated bytes.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
The current implementation relies on preprocessor concatenation to
work. This makes it incompatible with any content which expansion
is not a valid preprocessor token such as strings, pointers, etc.
and therefore limits its usefulness. Replace it with an implementation
that can cope with all cases.
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
This change implements qsort() for the minimal libc via Heapsort.
Heapsort time complexity is O(n log(n)) in the best, average,
and worst cases. It is O(1) in space complexity (i.e. sorts
in-place) and is iterative rather than recursive. Heapsort is
not stable (i.e. does not preserve order of identical elements).
On cortex-m0, this implementation occupies ~240 bytes.
Fixes#28896
Signed-off-by: Christopher Friedt <chrisfriedt@gmail.com>
This change adds `k_cycle_get_64()` on platforms that
support a 64-bit cycle counter.
The interface functions `arch_k_cycle_get_64()` and
`sys_clock_cycle_get_64()` are also introduced.
Fixes#39934
Signed-off-by: Christopher Friedt <chrisfriedt@gmail.com>
This lets the toolchain header files determine how to use "restrict"
instead of having that decision down in the minimal libc library.
Signed-off-by: Bradley Bolen <bbolen@lexmark.com>
Those are especially useful with hardware device registers.
They behave the same way as their Linux equivalent, with a much
simpler implementation for now.
Example:
#define REG_FIELD_A GENMASK(6, 0)
#define REG_FIELD_B BIT(7)
#define REG_FIELD_C GENMASK(15, 8)
#define REG_FIELD_D GENMASK(31, 16)
Get:
a = FIELD_GET(REG_FIELD_A, reg);
b = FIELD_GET(REG_FIELD_B, reg);
Set:
reg = FIELD_PREP(REG_FIELD_A, 1) |
FIELD_PREP(REG_FIELD_B, 0) |
FIELD_PREP(REG_FIELD_C, c) |
FIELD_PREP(REG_FIELD_D, 0x40);
Modify:
reg &= ~REG_FIELD_C;
reg |= FIELD_PREP(REG_FIELD_C, c);
Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
