Add -Wdouble-promotion and -Wfloat-conversion for most ports to ban out
implicit floating point conversions, and add extra Travis builds using
MICROPY_FLOAT_IMPL_FLOAT to uncover warnings which weren't found
previously. For the unix port -Wsign-comparison is added as well but only
there since only clang supports this but gcc doesn't.
Initially some of these were found building the unix coverage variant on
MacOS because that build uses clang and has -Wdouble-promotion enabled, and
clang performs more vigorous promotion checks than gcc. Additionally the
codebase has been compiled with clang and msvc (the latter with warning
level 3), and with MICROPY_FLOAT_IMPL_FLOAT to find the rest of the
conversions.
Fixes are implemented either as explicit casts, or by using the correct
type, or by using one of the utility functions to handle floating point
casting; these have been moved from nativeglue.c to the public API.
Now that error string compression is supported it's more important to have
consistent error string formatting (eg all lowercase English words,
consistent contractions). This commit cleans up some of the strings to
make them more consistent.
This macro is used to implement global serialisation, typically by
disabling IRQs. On the unix port, if threading is enabled, use the
existing thread mutex (that protects the thread list structure) for this
purpose. Other places in the code (eg the scheduler) assume this macro
will provide serialisation.
Based on eg 1e6fd9f2b4, it's understood that
the intention for unix builds is that regular builds disable assert, but
the coverage build should set -O0 and enable asserts.
It looks like this didn't work (even before variants were introduced, eg at
v1.11) -- coverage always built with -Os and -DNDEBUG.
This commit makes it possible for variants to have finer-grained control
over COPT flags, and enables assert() and -O0 on coverage builds.
Other variants already match the defaults so they have been updated.
Following up to 5e6cee07ab, some systems (eg
FreeBSD 12.0 64-bit) will crash if the stack-overflow margin is too small.
It seems the margin of 8192 bytes (or thereabouts) is always needed. This
commit adds this much margin if the requested stack size is too small.
Fixes issue #5824.
These were found by buiding the unix coverage variant on macOS (so clang
compiler). Mostly, these are fixing implicit cast of float/double to
mp_float_t which is one of those two and one mp_int_t to size_t fix for
good measure.
https://www.python.org/dev/peps/pep-0475/
This implements something similar to PEP 475 on the unix port, and for the
VfsPosix class.
There are a few differences from the CPython implementation:
- Since we call mp_handle_pending() between any ENITR's, additional
functions could be called if MICROPY_ENABLE_SCHEDULER is enabled, not
just signal handlers.
- CPython only handles signal on the main thread, so other threads will
raise InterruptedError instead of retrying. On MicroPython,
mp_handle_pending() will currently raise exceptions on any thread.
A new macro MP_HAL_RETRY_SYSCALL is introduced to reduce duplicated code
and ensure that all instances behave the same. This will also allow other
ports that use POSIX-like system calls (and use, eg, VfsPosix) to provide
their own implementation if needed.
The stack size adjustment for detecting stack overflow in threads was not
taking into account that the requested stack size could be <= 8k, in which
case the subtraction would overflow. This is fixed in this commit by
ensuring that the adjustment can't be more than the available size.
This fixes the test tests/thread/thread_stacksize1.py which sometimes
crashes with a segmentation fault because of an uncaught NLR jump, which is
a "maximum recursion depth exceeded" exception.
Suggested-by: @dpgeorge
This removes the port-specific definition of MP_PLAT_PRINT_STRN on the unix
port. Since fee7e5617f this is no longer a
single function call so we are not really optimising anything over using
the default definition of MP_PLAT_PRINT_STRN which calls
mp_hal_stdout_tx_strn_cooked().
This commit adds micropython.heap_locked() which returns the current
lock-depth of the heap, and can be used by Python code to check if the heap
is locked or not. This new function is configured via
MICROPY_PY_MICROPYTHON_HEAP_LOCKED and is disabled by default.
This commit also changes the return value of micropython.heap_unlock() so
it returns the current lock-depth as well.
sys.stdout.flush() is needed on CPython to flush the output, and the change
in this commit makes such an expression also work on MicroPython (although
MicroPython doesn't actual need to do any flushing).
This string is recognised by uncrustify, to disable formatting in the
region marked by these comments. This is necessary in the qstrdef*.h files
to prevent modification of the strings within the Q(...). In other places
it is used to prevent excessive reformatting that would make the code less
readable.
If the built-in input() is enabled (which it is by default) then it needs
some form of readline, so supply it with one when MICROPY_USE_READLINE=0.
Fixes issue #5658.
This changes the signal used to trigger garbage collection from SIGUSR1 to
SIGRTMIN + 5. SIGUSR1 is quite common compared to SIGRTMIN (measured by
google search results) and is more likely to conflict with libraries that
may use the same signal.
POSIX specifies that there are at least 8 real-time signal so 5 was chosen
as a "random" number to further avoid potential conflict with libraries
that may use SIGRTMIN or SIGRTMAX.
Also, if we ever have a `usignal` module, it would be nice to leave SIGUSR1
and SIGUSR2 free for user programs.
The install target is current broken when PROG is used to override the
default executable name. This fixes it by removing the redundant TARGET
variable and uses PROG directly instead.
The install and uninstall targets are also moved to the common unix
Makefile so that all variants can be installed in the same way.
Currently it is not possible to override PREFIX when installing micropython
using the makefile. It is common practice to be able to run something like
this:
$ make install PREFIX=/usr DESTDIR=/tmp/staging
This fixes such usage.
The mp_keyboard_interrupt() function does exactly what is needed here, and
using it gets ctrl-C working when MICROPY_ENABLE_SCHEDULER is enabled on
these ports (and MICROPY_ASYNC_KBD_INTR is disabled).
Pending exceptions would otherwise be handled later on where there may not
be an NLR handler in place.
A similar fix is also made to the unix port's REPL handler.
Fixes issues #4921 and #5488.
Previous behaviour is when this argument is set to "true", in which case
the function will raise any pending exception. Setting it to "false" will
cancel any pending exception.
CPython also has os.environ, which should be used instead of os.getenv()
due to caching in the os.environ mapping. But for MicroPython it makes
sense to only implement the basic underlying methods, ie getenv/putenv/
unsetenv.
This adds a -h option to print the usage help text and adds a new, shorter
error message that is printed when invalid arguments are given. This
behaviour follows CPython (and other tools) more closely.
This commit modifies the usage() function to only print the -v option help
text when MICROPY_DEBUG_PRINTERS is enabled. The -v option requires this
build option to be enabled for it to have any effect.
The usage text is also modified to show the -i and -m options, and also
show that running a command, module or file are mutually exclusive.
This adds support for a MICROPYINSPECT environment variable that works
exactly like PYTHONINSPECT; per CPython docs:
If this is set to a non-empty string it is equivalent to specifying the
-i option.
This variable can also be modified by Python code using os.environ to
force inspect mode on program termination.
When stdout is redirected it is useful to have errors printed to stderr
instead of being redirected.
mp_stderr_print() can't be used in these two instances since the
MicroPython runtime is not running so we use fprintf(stderr) instead.
This modifies the signature of mp_thread_set_state() to use
mp_state_thread_t* instead of void*. This matches the return type of
mp_thread_get_state(), which returns the same value.
`struct _mp_state_thread_t;` had to be moved before
`#include <mpthreadport.h>` since the stm32 port uses it in its
mpthreadport.h file.
It is not safe to enable MICROPY_ASYNC_KBD_INTR and MICROPY_PY_THREAD_GIL
at the same time. This will trigger a compiler error to ensure that it
is not possible to make this mistake.
Addition of GIL EXIT/ENTER pairs are:
- modos: release the GIL during system calls. CPython does this as well.
- moduselect: release the GIL during the poll() syscall. This call can be
blocking, so it is important to allow other threads to run at this time.
- modusocket: release the GIL during system calls. Many of these calls can
be blocking, so it is important to allow other threads to run.
- unix_mphal: release the GIL during the read and write syscalls in
mp_hal_stdin_rx_chr and mp_hal_stdout_tx_strn. If we don't do this
threads are blocked when the REPL or the builtin input function are used.
- file, main, mpconfigport.h: release GIL during syscalls in built-in
functions that could block.
When CFLAGS_EXTRA/LDFLAGS_EXTRA (or anything) is set on the command line of
a make invocation then it will completely override any setting or appending
of these variables in the makefile(s). This means builds like the coverage
variant will have their mpconfigvariant.mk settings overridden. Fix this
by using CFLAGS/LDFLAGS exclusively in the makefile(s), reserving the
CFLAGS_EXTRA/LDFLAGS_EXTRA variables for external command-line use only.
This commit adds backward-word, backward-kill-word, forward-word,
forward-kill-word sequences for the REPL, with bindings to Alt+F, Alt+B,
Alt+D and Alt+Backspace respectively. It is disabled by default and can be
enabled via MICROPY_REPL_EMACS_WORDS_MOVE.
Further enabling MICROPY_REPL_EMACS_EXTRA_WORDS_MOVE adds extra bindings
for these new sequences: Ctrl+Right, Ctrl+Left and Ctrl+W.
The features are enabled on unix micropython-coverage and micropython-dev.
Invoking "make" will still build the standard "micropython" executable, but
other variants are now build using, eg, "make VARIANT=minimal". This
follows how bare-metal ports specify a particular board, and allows running
any make target (eg clean, test) with any variant.
Convenience targets (eg "make coverage") are provided to retain the old
behaviour, at least for now.
See issue #3043.
Instances of the slice class are passed to __getitem__() on objects when
the user indexes them with a slice. In practice the majority of the time
(other than passing it on untouched) is to work out what the slice means in
the context of an array dimension of a particular length. Since Python 2.3
there has been a method on the slice class, indices(), that takes a
dimension length and returns the real start, stop and step, accounting for
missing or negative values in the slice spec. This commit implements such
a indices() method on the slice class.
It is configurable at compile-time via MICROPY_PY_BUILTINS_SLICE_INDICES,
disabled by default, enabled on unix, stm32 and esp32 ports.
This commit also adds new tests for slice indices and for slicing unicode
strings.
The existing uos.remove cannot be used to remove directories, instead
uos.rmdir is needed. And also provide uos.rename to get a good set of
filesystem functionality without requiring additional Python-level os
functions (eg using ffi).
This commit removes the Makefile-level MICROPY_FATFS config and moves the
MICROPY_VFS_FAT config to the Makefile level to replace it. It also moves
the include of the oofatfs source files in the build from each port to a
central place in extmod/extmod.mk.
For a port to enabled VFS FAT support it should now set MICROPY_VFS_FAT=1
at the level of the Makefile. This will include the relevant oofatfs files
in the build and set MICROPY_VFS_FAT=1 at the C (preprocessor) level.
When loading a manifest file, e.g. by include(), it will chdir first to the
directory of that manifest. This means that all file operations within a
manifest are relative to that manifest's location.
As a consequence of this, additional environment variables are needed to
find absolute paths, so the following are added: $(MPY_LIB_DIR),
$(PORT_DIR), $(BOARD_DIR). And rename $(MPY) to $(MPY_DIR) to be
consistent.
Existing manifests are updated to match.
This commit adds support for sys.settrace, allowing to install Python
handlers to trace execution of Python code. The interface follows CPython
as closely as possible. The feature is disabled by default and can be
enabled via MICROPY_PY_SYS_SETTRACE.
mp_compile no longer takes an emit_opt argument, rather this setting is now
provided by the global default_emit_opt variable.
Now, when -X emit=native is passed as a command-line option, the emitter
will be set for all compiled modules (included imports), not just the
top-level script.
In the future there could be a way to also set this variable from a script.
Fixes issue #4267.
Enabled via MICROPY_PY_URE_DEBUG, disabled by default (but enabled on unix
coverage build). This is a rarely used feature that costs a lot of code
(500-800 bytes flash). Debugging of regular expressions can be done
offline with other tools.
As per PEP 485, this function appeared in for Python 3.5. Configured via
MICROPY_PY_MATH_ISCLOSE which is disabled by default, but enabled for the
ports which already have MICROPY_PY_MATH_SPECIAL_FUNCTIONS enabled.
This allows figuring out the number of bytes in the memoryview object as
len(memview) * memview.itemsize.
The feature is enabled via MICROPY_PY_BUILTINS_MEMORYVIEW_ITEMSIZE and is
disabled by default.
The original code called setsockopt(SO_RCVTIMEO/SO_SNDTIMEO) with NULL
timeout structure argument, which is an illegal usage of that function.
The old code also didn't validate the return value of setsockopt, missing
the bug completely.
When building with link time optimization enabled it is possible both
gc_collect() and gc_collect_regs_and_stack() get inlined into gc_alloc()
which can result in the regs variable being pushed on the stack earlier
than some of the registers. Depending on the calling convention, those
registers might however contain pointers to blocks which have just been
allocated in the caller of gc_alloc(). Then those pointers end up higher on
the stack than regs, aren't marked by gc_collect_root() and hence get
sweeped, even though they're still in use.
As reported in #4652 this happened for in 32-bit msvc release builds:
mp_lexer_new() does two consecutive allocations and the latter triggered a
gc_collect() which would sweep the memory of the first allocation again.
As mentioned in #4450, `websocket` was experimental with a single intended
user, `webrepl`. Therefore, we'll make this change without a weak
link `websocket` -> `uwebsocket`.
If opening of /dev/mem has failed an `OSError` is appropriately raised, but
the next time `mem8/16/32` is accessed the invalid file descriptor is used
and the program gets a SIGSEGV.
Python defines warnings as belonging to categories, where category is a
warning type (descending from exception type). This is useful, as e.g.
allows to disable warnings selectively and provide user-defined warning
types. So, implement this in MicroPython, except that categories are
represented just with strings. However, enough hooks are left to implement
categories differently per-port (e.g. as types), without need to patch each
and every usage.
One can't use pthread calls in a signal handler because they are not
async-signal-safe (see man signal-safety). Instead, sem_post can be used
to post from within a signal handler and this should be more efficient than
using a busy wait loop, waiting on a volatile variable.
We standardized to provide uos.remove() as a more obvious and user-friendly
name. That's what written in the docs. The Unix port implementation
predates this convention, so update it now.
Configurable via MICROPY_MODULE_GETATTR, disabled by default. Among other
things __getattr__ for modules can help to build lazy loading / code
unloading at runtime.
Configurable via MICROPY_PY_BUILTINS_STR_COUNT. Default is enabled.
Disabled for bare-arm, minimal, unix-minimal and zephyr ports. Disabling
it saves 408 bytes on x86.
1. Return correct error code for non-blocking vs timed out socket
(POSIX returns EAGAIN for both, we want ETIMEDOUT in case of timed
out socket). To achieve this, blocking/non-blocking flag is added
to the mp_obj_socket_t, to avoid issuing fcntl() syscall each time
EAGAIN occurs. (mp_obj_socket_t used to be 8 bytes, having some room
in a standard 16-byte alloc block.)
2. Handle socket.settimeout(0) properly - in Python, that means
non-blocking mode, but SO_RCVTIMEO/SO_SNDTIMEO of 0 is infinite
timeout.
3. Overall, make sure that socket.settimeout() call switches blocking
state as expected.
This will allow to e.g. implement HTTP Digest authentication.
Adds 540 bytes for x86_32, 332 for arm_thumb2 (for Unix port, which already
includes axTLS library).
This commit adds the math.factorial function in two variants:
- squared difference, which is faster than the naive version, relatively
compact, and non-recursive;
- a mildly optimised recursive version, faster than the above one.
There are some more optimisations that could be done, but they tend to take
more code, and more storage space. The recursive version seems like a
sensible compromise.
The new function is disabled by default, and uses the non-optimised version
by default if it is enabled. The options are MICROPY_PY_MATH_FACTORIAL
and MICROPY_OPT_MATH_FACTORIAL.
Changes made:
- make use of MP_OBJ_TO_PTR and MP_OBJ_FROM_PTR where necessary
- fix shadowing of index variable i, renamed to j
- fix type of above variable to size_t to prevent comparison warning
- fix shadowing of res variable
- use "(void)" instead of "()" for functions that take no arguments
If DTTOIF() macro is not defined, the code refers to MP_S_IFDIR, etc.
symbols defined in extmod/vfs.h, so should include it.
This fixes build for Android.
"coverage" build uses different BUILD directory comparing to the normal
build. Previously, any build picked up libaxtls.a from normal build's
directory, but that was fixed recently. So, for each build, we must
build axtls explicitly.
This fixes Travis build in particular.
This patch in effect renames MICROPY_DEBUG_PRINTER_DEST to
MICROPY_DEBUG_PRINTER, moving its default definition from
lib/utils/printf.c to py/mpconfig.h to make it official and documented, and
makes this macro a pointer rather than the actual mp_print_t struct. This
is done to get consistency with MICROPY_ERROR_PRINTER, and provide this
macro for use outside just lib/utils/printf.c.
Ports are updated to use the new macro name.
This mechanism will scale to to an arbitrary number of pollable objects, so
long as they implement the MP_STREAM_GET_FILENO ioctl. Since ussl objects
pass through ioctl requests transparently to the underlying socket object,
it will allow ussl sockets to be polled. And a user object with uio.IOBase
as a base could support polling.
The API follows guidelines of https://www.python.org/dev/peps/pep-0272/,
but is optimized for code size, with the idea that full PEP 0272
compatibility can be added with a simple Python wrapper mode.
The naming of the module follows (u)hashlib pattern.
At the bare minimum, this module is expected to provide:
* AES128, ECB (i.e. "null") mode, encrypt only
Implementation in this commit is based on axTLS routines, and implements
following:
* AES 128 and 256
* ECB and CBC modes
* encrypt and decrypt
This behaviour of a NULL write C method on a stream that uses the write
adaptor objects is no longer supported. It was only ever used by the
coverage build for testing the fail path of mp_get_stream_raise().
Now that the coverage build has fully switched to the VFS sub-system these
functions were no longer available, so add them to the uos_vfs module.
Also, vfs_open is no longer needed, it's available as the built-in open.
The unix coverage build is now switched fully to the VFS implementation, ie
the uos module is the uos_vfs module. For example, one can now sandbox uPy
to their home directory via:
$ ./micropython_coverage
>>> import uos
>>> uos.umount('/') # unmount existing root VFS
>>> vfs = uos.VfsPosix('/home/user') # create new POSIX VFS
>>> uos.mount(vfs, '/') # mount new POSIX VFS at root
Some filesystem/OS features may no longer work with the coverage build due
to this change, and these need to be gradually fixed.
The standard unix port remains unchanged, it still uses the traditional uos
module which directly accesses the underlying host filesystem.
This patch moves the implementation of stream closure from a dedicated
method to the ioctl of the stream protocol, for each type that implements
closing. The benefits of this are:
1. Rounds out the stream ioctl function, which already includes flush,
seek and poll (among other things).
2. Makes calling mp_stream_close() on an object slightly more efficient
because it now no longer needs to lookup the close method and call it,
rather it just delegates straight to the ioctl function (if it exists).
3. Reduces code size and allows future types that implement the stream
protocol to be smaller because they don't need a dedicated close method.
Code size reduction is around 200 bytes smaller for x86 archs and around
30 bytes smaller for the bare-metal archs.
This test for calling gc_realloc() while the GC is locked can be done in
pure Python, so better to do it that way since it can then be tested on
more ports.
These new tests cover cases that can't be reached from Python and get
coverage of py/mpz.c to 100%.
These "unreachable from Python" pieces of code could be removed but they
form an integral part of the mpz C API and may be useful for non-Python
usage of mpz.
This function was implemented as an experiment, and was enabled only in
unix port. To remind, it allows to access arbitrary files frozen as
source modules (vs bytecode).
However, further experimentation showed that the same functionality can
be implemented with frozen bytecode. The process requires more steps, but
with suitable toolset it doesn't matter patch. This process is:
1. Convert binary files into "Python resource module" with
tools/mpy_bin2res.py.
2. Freeze as the bytecode.
3. Use micropython-lib's pkg_resources.resource_stream() to access it.
In other words, the extra step is using tools/mpy_bin2res.py (because
there would be wrapper for uio.resource_stream() anyway).
Going frozen bytecode route allows more flexibility, and same/additional
efficiency:
1. Frozen source support can be disabled altogether for additional code
savings.
2. Resources could be also accessed as a buffer, not just as a stream.
There're few caveats too:
1. It wasn't actually profiled the overhead of storing a resource in
"Python resource module" vs storing it directly, but it's assumed that
overhead is small.
2. The "efficiency" claim above applies to the case when resource
file is frozen as the bytecode. If it's not, it actually will take a
lot of RAM on loading. But in this case, the resource file should not
be used (i.e. generated) in the first place, and micropython-lib's
pkg_resources.resource_stream() implementation has the appropriate
fallback to read the raw files instead. This still poses some distribution
issues, e.g. to deployable to baremetal ports (which almost certainly
would require freezeing as the bytecode), a distribution package should
include the resource module. But for non-freezing deployment, presense
of resource module will lead to memory inefficiency.
All the discussion above reminds why uio.resource_stream() was implemented
in the first place - to address some of the issues above. However, since
then, frozen bytecode approach seems to prevail, so, while there're still
some issues to address with it, this change is being made.
This change saves 488 bytes for the unix x86_64 port.
This patch simplifies the str creation API to favour the common case of
creating a str object that is not forced to be interned. To force
interning of a new str the new mp_obj_new_str_via_qstr function is added,
and should only be used if warranted.
Apart from simplifying the mp_obj_new_str function (and making it have the
same signature as mp_obj_new_bytes), this patch also reduces code size by a
bit (-16 bytes for bare-arm and roughly -40 bytes on the bare-metal archs).
The SHA1 hashing functionality is provided via the "axtls" library's
implementation, and hence is unavailable when the "axtls" library is not being
used. This change provides the same SHA1 hashing functionality when using the
"mbedtls" library by using its implementation instead.
With inplace methods now disabled by default, it makes sense to enable
reverse methods, as they allow for more useful features, e.g. allow
for datetime module to implement both 2 * HOUR and HOUR * 2 (where
HOUR is e.g. timedelta object).
Unix naming is historical, before current conventions were established.
All other ports however have it as "modusocket.c", so rename for
consistency and to avoid confusion.
The uos.dupterm() signature and behaviour is updated to reflect the latest
enhancements in the docs. It has minor backwards incompatibility in that
it no longer accepts zero arguments.
The dupterm_rx helper function is moved from esp8266 to extmod and
generalised to support multiple dupterm slots.
A port can specify multiple slots by defining the MICROPY_PY_OS_DUPTERM
config macro to an integer, being the number of slots it wants to have;
0 means to disable the dupterm feature altogether.
The unix and esp8266 ports are updated to work with the new interface and
are otherwise unchanged with respect to functionality.
Header files that are considered internal to the py core and should not
normally be included directly are:
py/nlr.h - internal nlr configuration and declarations
py/bc0.h - contains bytecode macro definitions
py/runtime0.h - contains basic runtime enums
Instead, the top-level header files to include are one of:
py/obj.h - includes runtime0.h and defines everything to use the
mp_obj_t type
py/runtime.h - includes mpstate.h and hence nlr.h, obj.h, runtime0.h,
and defines everything to use the general runtime support functions
Additional, specific headers (eg py/objlist.h) can be included if needed.
This adds a new configuration option to print runtime warnings and errors to
stderr. On Unix, CPython prints warnings and unhandled exceptions to stderr,
so the unix port here is configured to use this option.
The unix port already printed unhandled exceptions on the main thread to
stderr. This patch fixes unhandled exceptions on other threads and warnings
(issue #2838) not printing on stderr.
Additionally, a couple tests needed to be fixed to handle this new behavior.
This is done by also capturing stderr when running tests.
Current users of fixed vstr buffers (building file paths) assume that there
is no overflow and do not check for overflow after building the vstr. This
has the potential to lead to NULL pointer dereferences
(when vstr_null_terminated_str returns NULL because it can't allocate RAM
for the terminating byte) and stat'ing and loading invalid path names (due
to the path being truncated). The safest and simplest thing to do in these
cases is just raise an exception if a write goes beyond the end of a fixed
vstr buffer, which is what this patch does. It also simplifies the vstr
code.
If, for class X, X.__add__(Y) doesn't exist (or returns NotImplemented),
try Y.__radd__(X) instead.
This patch could be simpler, but requires undoing operand swap and
operation switch to get non-confusing error message in case __radd__
doesn't exist.
This is to keep the top-level directory clean, to make it clear what is
core and what is a port, and to allow the repository to grow with new ports
in a sustainable way.
