The handler of the microkernel events are not declared properly.
This causes the public event objects to complain about
undefined functions. Fix this by declaring the function
before using it.
JIRA: ZEP-85
Change-Id: I68afe777bad66e010c92e637806a48911b2f52f4
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Directory layout per SDK 0.8 structure. Using toplevel libs for
now.
Change-Id: I7852a9def09fb0a5f0a6e496249d2ebb86720d8f
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Instead of defining the output format in the toolchain environment
setup, define this per SoC that supports the ABI.
This will allow us to directly referencing an installed toolchain
without the need for declaring ZEPHYR_GCC_VARIANT
Change-Id: Icd1a7a21acbf075a51854fb2ebb3ef06788ce9b5
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
It allows to silently ignore non-existent files.
When using wildcards in Kconfig source files
(e.g. source folder/*/Kconfig), it is possible to refer files
that does not exist. In the previous example, it is possible
that Kconfig files do not exist in one of the folder's
subfolders and it is not a requirement for the file to exist
in each one of the subfolders.
Additionally, it fixes an issue for wildcards in the file name.
If the name contains a wildcard, Kconfig should iterate
over each file included in the wildcard
(e.g source folder/myKconfig.*)
Jira: ZEP-177
Change-Id: I5aa192ca1e2df83461b12a86fe29a98cd95c4256
Signed-off-by: Juan Manuel Cruz <juan.m.cruz.alcaraz@intel.com>
Many sub-systems might require to set a callback on different pins.
Thus enabling it via changing the API.
It is also possible to retrieve private-data in the callback handler
using CONTAINER_OF() macro (include/misc/util.h).
Former API is still available, and is emulated through the new one.
Using both should not be a problem as it's using new API calls.
However, it's now better to start using the new API.
Change-Id: Id16594202905976cc524775d1cd3592b54a84514
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
generates lots of false postives, so make it warn only.
Change-Id: I0b2fedc564c29bff32f7c48702a9ad54a969650b
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
() The test names are getting longer so adds more padding.
() Aligns the numerical size numbers for easier visual
comparison when scrolling in shell.
Change-Id: Id218a641e9d5dffb77bf813ffbd828c48d14020f
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Some boards require a newer version of openocd, this will
allow overriding the default version that comes with the SDK.
Change-Id: I70113182c3a2c8610e3b36542c6057ff3f0df547
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The ordering of the list of platforms is in priority order,
as the sanitycheck --platform-limit option chooses the first
N platforms in this list when doing a run. Move arduino_101 to
second position. qemu_x86_iamcu is still top priority as it
can execute code as part of the tests.
Change-Id: I989d91c66a0a741ce8eebe6279ee4512c33eb5e9
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Add the last path information in the testcase root to the
test name, so that there is much less of chance of name
collisions and the reports come out with the same names
as before.
Change-Id: I75ed6ee96251058dd2547e57be31db3c2517a497
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
A recent tree change moved testcases to both the tests/ and
samples/ directory. The default testcase root was just made
the root of the Zephyr tree, which could lead to potentially
long scan times as every single directory in-tree was checked.
For example, it could be very time-consuming to scan everything
under the sanity-out/ dir if -n is used and --all was used in
a previous run.
Now we allow multiple testcase roots to be supplied on the
command line, and by default only scan under tests/ and
samples/.
Change-Id: I8dac747835b87801474e08c236913e04db64f87b
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This adds the ability for sanitycheck to use arbitrary report file
for size comparison. This is useful when optimizing binary size.
For example, a report can be generated with master head, and used
as base for comparison.
Change-Id: Id8c61146c498ba62c4cc77a9ef7d390c0fb11487
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Instead of running nm/objdump and parsing the results one testcase
at a time, do these in parallel to speed up sanity check a bit.
The sanitycheck now requires at least Python 3.2 to run because of
the usage of concurrent.futures.
This reduces build time (--all -b) from 4:30 to 3:08 on 24C/48T
Xeon-E5.
Change-Id: I8e7c1efb2f473c7f2b65658f8ed9a101ed091eea
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
The "initlevel" section is being metioned twice. It does not affect
the final size calculation, but just remove it just for reader's
sanity.
Change-Id: I0e7d6e82c730631d2c83ca09dde0f675805840f3
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
The script will now run under either Python 2 or Python 3.
Change-Id: Ia7e6647b2331ff9edfbdec7b7357439f7095a3bc
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
There are two major issues with the kconfig:
() Some of the config options have incorrect dependencies inside help
under menuconfig. For example, CONFIG_GPIO depends on BOARD_GALILEO.
() Since the SoC and board specific kconfig files are parsed first,
the help screen would say, for example, CONFIG_SPI is defined at
arch/arm/soc/fsl_frdm_k64f/Kconfig. This is incorrect because
the actual config is defined in drivers/spi/Kconfig.
These cause great confusion to users of menuconfig/xconfig.
To fix these, the SoC and board defaults are now to be parsed last.
Note that the position swapping of defaults in this patch is due to
the fact the the default parsed last will be used.
And, spi_test is broken due to the fact that it requires
CONFIG_SPI_INTEL_PORT_1, but never enables it anywhere. This is
bypassed for now.
Origin: refactored and edited from existing files
Change-Id: I2a4b1ae5be4d27e68c960aa47d91ef350f2d500f
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This is in preparation to move SoC and board kconfig defaults
to be parsed later.
Change-Id: If24bdf310dac7034da63f34c0e2add173fe75844
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This adds a debug option to print defaults and their conditions
inside config description under menuconfig/xconfig. This aids in
debugging of kconfig creation.
Change-Id: Ie651435f6b2115dac1bd3a6f7a1b3c2df3c9b0ed
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
If the arch ini file doesn't have a specific block for the platform,
just assume all defaults.
Change-Id: Id6ab1ede8936bdd6006a5c4fd2f9a4f6aaf42478
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
It is an unfortunate fact of life that our CI machines are often
severly underpowered virtual machines that take quite some time
to run sanitycheck --all. The only other option was to just use
the default test case filtering semantics which chose one platform
per arch to test.
Now it is possible to specify N platforms per arch with the new
--platform-limit option, allowing greater flexibility on how many
tests will be run.
The counter-intuitive use of '--platform all' or '--platform default'
is no longer supported as these use-cases are covered by other
command line options.
Change-Id: I45b0050df52e1a22e75534a3876d89e3c77698c9
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Rationale for defaulting to IAMCU variant is that our most
interesting x86 boards use this calling convention. --all
will test both.
Change-Id: Idaddfec51a74ae85691d348adeeb88a8528060ea
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
For some strange reason IAMCU sets e_machine to 0x06 which causes
QEMU to freak out. This is just x86 code with a different C
calling convention, hack it back to 0x03 (EM_386) before running
under the emulator.
Change-Id: Ia5d51b771cad41f3013eb3d6a17912c8909c9bac
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Restructure the kernel's object tracing implementation
to provide a public API that allows debug tools to use
the debug hooks easier and allows kernel developers
to extend the kernel's object tracing scope and include
new kernel objects easier.
The API provides the trace list abstraction to keep track
of different types of kernel objects. The API contains
a simple single-linked list implementation that allows
to save space and simplifies the access to the data for
debug tools such as gdb.
Change-Id: Ic4d393d584576f67f2c5b706e61bae08869debba
Signed-off-by: Juan Manuel Cruz <juan.m.cruz.alcaraz@intel.com>
Specifying HEAP_SIZE keyword in an app's MDEF file, results in
creating a new memory pool, which can be accessed using the
task_malloc() and task_free() APIs, which have the usual malloc/free
like semantics.
Expected format in MDEF file
HEAP_SIZE <value>
Change-Id: I0569cffeecf8a2c23c20c7b359256123ece91982
Signed-off-by: Jithu Joseph <jithu.joseph@intel.com>
Now if you try to call IRQ_CONNECT() multiple times on the same
IRQ line it will correcly fail the build, as intended.
Change-Id: Ia629ae3dd009bd3a2e1dbd9797dc2b723b02d234
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
These don't have necessary symbol information to determine if the
kernel is XIP or not. Fail instead of giving bogus data.
Change-Id: I87f6eeb5983f5275929e5b8d448a054b83e23d8f
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Python 2 isn't supported anymore and doesn't get fixes/updates.
For the long-term maintenance health of sanitycheck, convert to
use Python 3 interpreter, which mostly involved changes to how
strings are handled.
Change-Id: Ic42f5b2328001f04d876fd650986c4433968a76b
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Fix issue where a space is required after a comma:
-:245: ERROR:SPACING: space required after that ',' (ctx:VxB)
+ SECTION_PROLOGUE(_NOINIT_DMA_SECTION_NAME, (NOLOAD OPTIONAL),)
Now we will ignore such errors if the comma is followed by a ).
Change-Id: Ib38161cd11e650764c319ba373a1f7c1a6d80df6
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Update to commit 6b10df4257367dd0ead49f88df473972c00a8b5c from the Linux
kernel, relevant changes since last update:
- checkpatch: fix a number of COMPLEX_MACRO false positives
- checkpatch: improve macros with flow control test
- checkpatch: warn when casting constants to c90 int or longer types
- checkpatch: improve the unnecessary initialisers tests
- checkpatch: improve tests for fixes:, long lines and stack dumps in commit log
Change-Id: Ic4f8d925cd7c076e18eb2f2841913be61239aac2
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
This board is not supported and not available for general public.
Use the Quark SE CRB/Devboard instead.
Change-Id: Id0f8c08bbacb812ef00fe9502b4acecf4f31ffd7
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
We have moved tests around, update the files to get accurate footprint
changes per test case.
Change-Id: I563acef4665cb0ab6d3e7e01050a3a49f98b8b9d
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
MinGW have small conventions differences in the
call of some APIs. The following differences are
addressed:
- mkdir method parameters
- printf format for 64 bit integers
- utsname structure
- glob library
MinGW does not provide glob library.
This patch adapts the code to avoid this library when
building for MinGW on windows systems.
The new routines allow windows to support wildcards
(*, ?) on Kconfig paths.
zconf.lex.c_shipped is the file that the build system
compiles by default. If the user defines the
REGENERATE_PARSES environment variable for the build,
then zconf.lex.c_shipped is generated from the file
zconf.l. Both files are provided in the patch
to keep coherency with the REGENERATE_PARSES option.
Change-Id: I5b6ad24ead0521913ab6ea9da93551edcd2dc66b
Signed-off-by: Louise Mendoza <yonattan.a.louise.mendoza@intel.com>
Signed-off-by: Juan Manuel Cruz <juan.m.cruz.alcaraz@intel.com>
Specify the kernel being tested instead of relying on parsing
Makefile for KERNEL_TYPE.
This is needed for testcases that build for multiple kernels.
Change-Id: Ic4df2fa66eee4cd498955aee2b0da80a99d35e30
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
We have test cases in both samples and tests directories,
so cover both by scannining the root directory.
Change-Id: I5258d11b23ced2cc4fc88af4d40269310442dc9f
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The build system now pulls in the correct intrinsics libraries
when building for Cortex-M CPUs.
Change-Id: I55ed249cdeb04de8843ad4cdc0099e7031e368b0
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Some tests take a very long time to complete, add a directive in
testcase.ini to mark them as such. Unless --enable-slow is passed
on the command line, these tests are not executed, only compiled.
app_kernel and test_sha256 marked as slow test cases. It appears
test_sha256 was being skipped earlier due to platform_whitelist line
which has been removed; the test should be able to run on all
microkernel-supporting boards. Same with test_aes, which completes
very quickly and is not marked as 'slow'.
Change-Id: I39ec8212f3fa10122ff786c10b6659d22bae64e3
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This enables sanitycheck to build for Arduino Due.
Note that there is no on-board bluetooth so bluetooth tests
are skipped.
Change-Id: I4f564e68ec65f01b7b33a32f6d0d7749290c0e5b
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
