- `-fno-builtin` should not be necessary: a) by default, we assume
that a libc exists on the system the project is compiled for, but
b) even if the assumption does not hold, `-fno-builtin` can be
passed as an external CFLAG to the build system, plus c) the use
of builtins is expected to help both code size and performance.
(Except when jerry-math is enabled.)
- `-fno-stack-protector` should not be necessary: it is the default
not to instrument code with stack protection.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Determine the length of strings as early as possible and use that
to prevent unnecessary memory allocations, and in memory copies
instead of string concatenations.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Remove redundancy between all-in-one and all-in-one-source builds
by keeping only the second, and adopt the more established term
"amalgamated" build for it. This change includes the following:
- Replace `ENABLE_ALL_IN_ONE` and `ENABLE_ALL_IN_ONE_SOURCE` cmake
options with `ENABLE_AMALGAM` top-level option.
- Replace `--all-in-one` option of `build.py` helper with
`--amalgam`.
- Merge the `srcmerger.py` and `srcgenerator.py` tool scripts into
`amalgam.py` (with improvements).
- Update documentation.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Travis CI became very slow lately. Most of the jobs have already
been moved to GH Actions. This commit finishes the migration.
- IRC notification is adapted (result of jobs is not reported).
- Sonar integration is dropped without replacement, as its GH
Action does not support C projects.
- Coverity Scan integration is dropped without replacement as it
has no GH Action at all.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This allows users to specify a build type other than `MinSizeRel`
or `Debug` for `tools/build.py`. (Build type was already freely
configurable by invoking cmake directly with a custom
`CMAKE_BUILD_TYPE` option.)
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The pkgconfig specified `include/jerry-math` as the include
directory, but cmake installs header to `include/jerryscript-math`.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Make jobs run on latest versions of the OS's (except for
"Checks", which relies on specific versions of checkers coming
with Ubuntu 18.04)
- Bump actions/checkout to v2
- Rewrite `apt-get` to `apt` (almost the same, but `apt` is the
2nd gen tool)
- Drop step names (they took too much vertical space, didn't add
too much useful information, and were very typo-prone)
- Move superfluous environment variables (e.g., `BUILD_OPTIONS` or
`SKIP_LIST`) to their expansion sites (use YAML block scalars
`>-` where appropriate to enhance readability)
- Fix duplicate installation of pylint in the "Checks" job
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Normally, it is more usual and safe to use a toolchain's native
math library. Especially, if multiple components of a project use
math functions, in which case all components should be linked
against the same libm.
The libjerry-math can be used, of course, but as it needs extra
care and consideration, it should be opt-in.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
That "libm" in the name of the library resulted in awkward naming
on *nix systems (`libjerry-libm.*`, "lib" occurring twice). And the
name of the corresponding header is `math.h` anyway.
Note that this is a breaking change in some sense. The commit
contains no API change, but the build system does change for users
of the math library.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This can be used at configuration time to
- display version in diagnostics, and
- to write proper version number in pkgconfig files.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The minimal variant became quite meaningless lately. There were two
port APIs originally that had extra functions in the default port
in addition to the core-mandated implementations: the I/O and
Termination port APIs. However, the extra Termination API code was
removed a year ago, leaving some minimal extension in the I/O port
only. As the overhead of the extension is negligible, it is not
worth maintaining two library variants.
Therefore
- this commit removes the minimal variant of the default port lib,
- rewrites uses of the minimal variant to use the variant with the
I/O extension, and
- updates targets where I/O port code was copy-n-pasted.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
All installed headers and header directories of the project have
the `jerryscript-` prefix, except the math library, which installed
its `math.h` under `jerry-libm` directory.
This commit changes the name of the include directory to
`jerryscript-libm` to align it with the rest of the project.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Overflows in conversions from floating-point to integer are
undefined behavior in the C99 standard. (Clause 6.3.1.4: "If the
value of the integral part cannot be represented by the integer
type, the behavior is undefined.")
When UBSAN is enabled, this gets reported at `srand()` calls. (The
random seed is usually initialized using the date port API, which
represents dates as `double`s. But `srand` takes an `unsigned int`.
A simple cast from `double` to `unsigned` becomes undefined
behavior if the value is too large. And "now" is too large
nowadays. So, effectively, all executions start with an undefined
behavior.)
This patch fixes this by casting the floating-point value of the
date to an integer through a union.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Provide contributors with a checklist to ensure quality PRs and to
avoid common pitfalls. The items in the checklist are mostly beyond
the capabilities of automated checker tools and more like reminders
of best practices. Still, they can save lot of valuable reviewer
time (hopefully).
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Help both the community and ourselves by describing what
information is required in a bug report.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This reduces the noise when a user asks for help via `--help`, as
libfuzzer support is very developer-only.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Not running enqueued jobs (resulting from promises) caused false
alarms (internal assertion failures) at cleanup.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Allow command line tools to build together with libfuzzer driver.
Compile everything with `-fsanitize=fuzzer-no-link` to prevent
linking in libfuzzers's `main` symbol in all executables (causing
duplicate symbol errors in command line tools), and add
`-fsanitize=fuzzer` to the libfuzzer driver only.
- Make ASan optional when building with libfuzzer to allow the user
to choose freely from available sanitizers (e.g., UBSan, MSan,
HWASan).
- Stabilize libfuzzer by resetting PRNG seed at every invocation.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Current build test uses and documentation refers to release 1.1,
however, 2.0 has been out since Oct 2018. This commit bumps the
target to that latest release.
Co-authored-by: Roland Takacs <[email protected]>
JerryScript-DCO-1.0-Signed-off-by: Roland Takacs [email protected]
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Should core, ext, libm, and/or port libraries be properly installed
on some system, help compilation and linking against them by
providing standard `.pc` files, which can be picked up by
pkg-config.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This fixes potential out-of-bounds reads in jerry-ext when dealing
with integer argument mappings.
Fixes#2713
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The git hash mbed-os.lib was pointing to marked Mbed OS 5.5.5 and
was more than a year old and more than 10.000 commits behind latest
mbed-os master.
This commit forwards the reference to the latest Mbed OS release
(5.11.1).
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Both documentation and travis makefile were referring to NuttX 7.22,
which was released in Sep 2017. This commit bumps the version to the
latest release (7.27, from Nov 2018).
Additionally, it also sets the stlink tool to a fixed (and tagged)
version.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Until now, the ESP8266 target used a non-tagged git hash revision
of the RTOS SDK (from Oct 2017). Moreover, the compiler was also a
custom-built gcc.
This commit sets the RTOS SDK version to v2.1.0, released Oct 2018,
which is the latest old-style SDK for ESP. The commit also changes
the used gcc to an Espressif-provided pre-built toolchain, which is
tagged for the (latest) v3.0.1 SDK release.
(The SDK bump is not to the latest SDK version because the new
v3.x line of the SDK is "ESP-IDF style", which is incompatible with
the current target code base. On the other hand, the toolchain
comes from the latest SDK release because the v2.x line had no
pre-built toolchain suggestions -- but it builds v2.x sources
correctly.)
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- `sudo: false` has been deprecated.
- `install` step is skipped by default for C projects, so there is
no need setting it explicitly.
- Homebrew addon can be used to install dependencies for OSX jobs
(and, by experience, it is a lot faster than manually invoking
brew).
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- `ERR_SYSCALL` is a legacy that has not been used anymore since
the removal of jerry-libc. This commit removes it from the port
API.
- `ERR_DISABLED_BYTE_CODE` is a (relatively) recent addition to the
fatal code enum. At the time it was added, the documentation has
not been updated. This commit adds the missing documentation.
(Plus, it removes a superfluous `JERRY_UNREACHABLE` from after a
`jerry_fatal (ERR_DISABLED_BYTE_CODE)`.)
Note: As the port API is modified, this is an API-breaking change.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The format `1.0.{build}` is artificial, the 1.0 prefix has no
meaning, `{build}` is enough.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Recent failures of the Mbed OS target came from two independent
issues:
- some python packages started requiring Python>=2.7.10, and
- the latest Intervaltree 3.0.0 release was released broken.
This patch ensures a fresh python by 'lying' to the CI that this
is a Python project (Python images on the CI come with a recent
interpreter, not with the one shipped as default with the OS), and
locks Intervaltree to a stable version.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Since recently, the newly released mbed-host-tests 1.4.4 gets
installed during the python requirement installation step. That
package requires pyocd>=0.14.0, which cannot be satisfied for
whatever reason. This quick fix pins mbed-host-tests to version
1.4.2, which does not have unsatisfiable requirements.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The test runner used `&>>` to channel stderr to stdout and append
all that to a file. However, that's Bash 4 syntax, which is not
available everywhere, e.g., on MacOS. Bash pre-4 syntax requires
the classic `>>file 2>&1` notation.
This commit rewrites the test runner to use the portable syntax.
Moreover, it also rewrites `&>` to `>file 2>&1` for the sake of
consistency.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Because these feature guards don't have their counterpart options in cmake, they
seem to be less tested. This commit makes them build-tested at least.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Use builtin support of Travis CI for job names instead of using
a custom environment variable for the same purpose.
- Use `skip` in `script` and `install` steps instead of `true`.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The America/Los_Angeles time zone is already enforced for test262
Travis CI jobs but that doesn't guarantee the correctness of
locally executed tests. So, this patch moves the setting of the
`TZ` environment variable from `.travis.yml` to `run-tests.py`.
The date-related tests in the jerry test suite also rely on a time
zone (UTC). Thus, `TZ` is forced for those tests, too.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The only such test was a legacy of the long removed compact
profile, which is also removed.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Allow developers to choose whether they want a hard or a soft
implementation behind `assert`, i.e., whether they want to kill
the whole engine in case of an assertion failure in the executed
script or just throw a JS error.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
... not from the print handler extension. The sending of logs is
already in the port implementation.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Their length (size) is known at compile time. Therefore `sizeof`
is more efficient for them.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
It was an unused and undocumented macro that enabled diverting the
`jerry_port_get_current_context` calls, an approach that was not
used anywhere else in the Port API.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Under the cover of the debugger transport layer, allocation on the
engine's heap has been made available to the public. As there are
actually no restrictions on what the allocated memory can be used
for, the memory management functions better fit in the core part
of the API.
Closes#1805
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
There was quite some confusion about terminology around instances
and contexts. All the docs mentioned external contexts but
functions and types were referring to instances, and the relation
between these two concepts were not clear. This commit keeps
(external) context as the only surviving concept.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Get the macro definition indentations right.
- Define some of the macros only if system allocator is not in use.
- Ensure that macros with arguments don't cause unused variable
warnings even if memory statistics is disabled.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Moved global context's `JMEM_HEAP_SIZE` to jcontext.h as external
context's heap size is declared there, too.
- Moved the assert on `jmem_heap_t` vs `JMEM_HEAP_SIZE` to jcontext.c,
as both entities are declared in the corresponding header.
- Removed superfluous checks on `JMEM_HEAP_SIZE` as it is not a
publicly configurable macro (opposed to `CONFIG_MEM_HEAP_AREA_SIZE`).
- Ensured that all definitions of and references to `jmem_heap_t`,
`JERRY_HEAP_CONTEXT`, `JERRY_HEAP_SIZE`, and `JERRY_HEAP_AREA_SIZE`
are guarded by `#ifndef JERRY_SYSTEM_ALLOCATOR`, as they are
meaningless if the system allocator is used.
The commit also contains some stylistic changes in jcontext.h
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
It is superfluous to maintain multiple globals when the whole
reason of context is to keep them in a single place. It also
simplifies initialization and external context creation a bit.
Also removed unused lcache header includes.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The `[jerry|ecma]_char_ptr_t` types are some old legacy that are
used quite inconsistently. Their `[jerry|ecma]_char_t *` variants
are used a lot more often, so it's better to stick to one form
throughout the code base.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Since the removal of jerry-libc, only jerry-ext uses the
`FEATURE_INIT_FINI` optional feature. Thus, it's better to move it
from the global cmakelists to that of jerry-ext.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This means cross building with clang and using short enums. Also
bump RIOT OS version to latest 2018.07.
Note: On Travis CI, clang-3.9 is used for testing. That version is
widely available, from Ubuntu 14.04 to 18.04.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This will help to keep the debugger documentation up to date.
(Note: only compilation & linking is possible as execution would
require a connecting debugger client, which is not supported for
unit/doctests.)
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Rationale:
- There is no port under targets/ that would use it. All of them
turn it off when building.
- That's no surprise, as jerry-libc supports no barebone MCUs but
posix targets with syscalls only. Actually, that's Linux only,
because macOS builds have turned off the use of jerry-libc a
while ago.
- And there is no point in maintaining a highly restricted set of
libc functions: as soon as someone wants to use JerryScript in a
scenario that needs more functions than jerry-main, they have to
choose a different libc (most problably the compiler's default
one).
I think that we should not keep supporting an otherwise unused
library for the purposes of jerry-main on arm/x86/x64-linux only.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The documentation has not been updated when the HTML client got
removed nor when the new transport layer replaced the previous
debug server initialization process.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The patch also ensures that all components access only the public
headers of other components (except for unit tests, which are
allowed to use private headers, too).
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
This helps to avoid the use of non-public headers and
protocol-internal constants in external code (e.g., in jerry-port
and jerry-ext).
The patch also cleans up the necessary includes in jerry-core public
headers, and the include order in jerry-port/default public headers.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The distinction between public and private include directories was
not maintained and the install paths got confusing.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Clang 6 is more picky about implicit conversions and complains
about loss of presicion and/or change of signedness.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The renamed header doesn't contain any declarations related to
debugger-ws.c but contains the declaration of a function from
debugger-sha1.c.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
`FILE_PATTERNS` is a space-separated list, in contrary to what is
suggested by its documentation. The pattern `*.h, *.c` does not
match header files but files with `.h,` extension only. Rewriting
the current comma-separated pattern makes Doxygen actually process
header files. However, it also reveals several hitherto hidden
issues (mostly missing documentation) in the code. This patch fixes
some of these documentation problems (and lists the files with
outstanding issues in a 'backlog').
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Storing the error output of doxygen into a shell variable first and
echoing it later is error prone. In case of a long error output,
the whole shell will crash before being able to print the doxygen
diagnostics at all. This patch rewrites the check script to tee the
diagnostics (pipe them to console and store them to a file at the
same time) and check the output file size at the end.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
The progress messages of cppcheck produce a lengthy and verbose
log, making errors and warnings hard to find. This patch adds
`--quiet` to the invocation of `cppcheck`
Additionally, the patch relayouts the script to use a consistent
two-spaces indentation everywhere.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
There are some leftover global functions in the code that are not
referenced at all anymore. These functions are removed by this
patch.
There are also some global functions that are only used in their
own modules. These functions are made static by this patch.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
They are no big differences between the two forms as "" falls back
to <>. There are some inconsistencies in the code, though, which
are fixed by this patch.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
* Rewritten 128-bit integer "type" to use two 64-bit ints (they are
enough), and made it a proper struct instead of an array.
* Rewritten all single-bit shift loops to multi-bit shifts with the
help of CLZ, and used `__builtin_clzll` where available.
* Simplified (and documented) 128-bit DIV10 operation.
* Renamed 128-bit integer handling macros to use simpler names.
(And removed unused macros that were laying around.)
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
In `tools/build.py`:
- For the sake of readability, group CLI arguments as general build
options, options to control the building of components, and
component-specific options.
- To prevent duplications, remove the defaults from those CLI
arguments that correspond to CMake options and have defaults in
any of the CMakeLists. Should any of the defaults change, they
will have to be changed at a single place only. (Those options
that are not set on the command line of `tools/build.py` are not
passed as options to `cmake` either.)
- Convert `--unittests` and `--doctests` to ON/OFF options like the
rest of the component switches.
- Touch on some of the help messages of the CLI arguments.
Other changes:
- The change in `--unittests` and `--doctests` is a slightly CLI-
breaking change of `tools/build.py`. Thus, follow up on this in
`tools/run-tests.py`.
- Move `ENABLE_ALL_IN_ONE` into `jerry-core` as it is not a general
option but specific to that component.
- Remove the forcing of `ENABLE_ALL_IN_ONE` for some compilers/
platforms as it is still an option, not a hard requirement.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Add `--install` option to `tools/build.py`.
- Make use of `--install` in `tools/run-tests.py` by testing the
installed the executables instead of those in the build tree.
Related changes:
- Collect unit test binaries in the `tests` subdir of the build
tree instead of `bin`.
- The `ls`-based collection of the unit test binaries had some
shortcomings hitherto unrevealed (it didn't filter for files so
it could potentially "collect" dictionaries, too), which has now
been replaced with a more stable `find`-based solution.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Recent changes to master have already introduced API-breaking
changes. So, if someone is still including the legacy headers,
following up on the header file name changes will be the least of
their problems.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
As a result of earlier developments, there remained no other
instatiations of the `ALLOC`, `DEALLOC`, and `DECLARE_ROUTINES_FOR`
templates. So, it's simpler to write the alloc/dealloc routines
directly for `ecma_number_t` as well.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
- Code contains syntactic duplications (e.g., build options copy-
pasted repeatedly).
- Build options for test suites contain semantic duplications
(ES5.1 profile builds happen multiple times, once by not
specifying a profile and once by specifying es5.1 profile
explicitly).
- External build options are not taken into account when detecting
duplicated builds.
This patch provides improvement for these issues.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
None of the code paths have been tested for long and especially the
Freya code paths have been long abandoned. This patch:
- Merges Valgrind-Freya into Valgrind code path (there should be no
need to choose between them, Valgrind should work just fine).
- Removes leftover code (`VALGRIND_FREYA_CHECK_MEMPOOL_REQUEST` and
`valgrind_freya_mempool_request`).
- Adds `JMEM_` prefix to Valgrind-related macros (to correctly
leave the `VALGRIND_` prefix to Valgrind).
- Moves the definition of the Valgrind-related macros to a common
header to avoid duplication.
Note: Adding a CI job to perform Valgrind Memchecks is left for
follow-up as it turns out to be excessively slow (>50 mins for a
`--jerry-tests --jerry-test-suite` run, and even a simple
`--jerry-tests` may get terminated prematurely because of timeout
issues).
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
If the result of a build is not directly executable on the host
system or needs an execution runtime for any other reason (e.g.,
cross-compiled ARM binaries on Intel, emscripten-generated JS
files, binaries built with Valgrind support) then the current test
runners cannot work with them. This patch makes test runners
execute binaries via a runtime given in the `$RUNTIME` environment
variable (if it is specified).
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
`JERRY_UNREACHABLE`s often signal code structure that could be
improved: they can usually either be rewritten to `JERRY_ASSERT`s
or eliminated by restructuring loops, `if`s or `#if`s. Roughly,
the only valid occurences are in default cases of `switch`es. And
even they can often be merged into non-default cases.
Moreover, it is dangerous to write meaningful code after
`JERRY_UNREACHABLE` because it pretends as if there was a way to
recover from an impossible situation.
This patch rewrites/eliminates `JERRY_UNREACHABLE`s where possible
and removes misleading code from after them.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss [email protected]
Akos Kiss