GitLab

Printing of uPy floats can differ by the floating-point precision on
different architectures (eg 64-bit vs 32-bit x86), so it's not possible to
using printing of floats in some parts of this test.  Instead we can just
check for equivalence with what is known to be the correct answer.

Some older compilers cannot deduce that q_last is always written to before
being read.

For cases where size_t is smaller than mp_int_t (eg nan-boxing builds) the
difference between two size_t's is not sign extended into mp_int_t and so
the result is never negative.  This patch fixes this bug by using ssize_t
for the type of the result.

If MICROPY_PY_ALL_SPECIAL_METHODS is enabled then dir() will now delegate
to the special method __dir__ if the object it is listing has this method.

This gives dir() better behaviour when listing the attributes of a user
type that defines __getattr__: it will now not list those attributes for
which __getattr__ raises AttributeError (meaning the attribute is not
supported by the object).

This patch fixes the possibility of a crash of the REPL when tab-completing
an object which raises an exception when its attributes are accessed.

See issue #3729.

It now allows __getattr__ in a user type to raise AttributeError when the
attribute does not exist.

This new helper function acts like mp_load_method_maybe but is wrapped in
an NLR handler so it can catch exceptions.  It prevents AttributeError from
propagating out, and optionally all other exceptions.  This helper can be
used to fully implement hasattr (see follow-up commit), and also for cases
where mp_load_method_maybe is used but it must now raise an exception.

The list of qstrs starts with MP_QSTR_NULL followed by MP_QSTR_, and these
should never appear in dir() or REPL tab completion, so skip them.

Commit e269cabe added a check that the
first argument to the to_bytes() method is an integer, and now uPy
follows CPython behaviour and raises a TypeError for this test.

Note: CPython checks the argument types before checking the number of
arguments, but uPy does it the other way around, so they give different
exception messages for this test, but still the same type, a TypeError.

Commit 68c28174 implemented checking for
valid utf-8 data.

Commit 1e70fda6 fixes this difference.

On this 32-bit arch there's no need to use the long version of the format
specifier.  It's only there to appease the compiler which checks the type
of the args passed to printf.  Removing the "l" saves a bit of code space.

Using MP_STREAM_POLL_HUP for ERR_RST state follows how *nix handles this
case.

The order of function calls in an arithmetic expression is undefined and so
they must be written out as sequential statements.

Thanks to @dv-extrarius for reporting this issue, see issue #3690.

If a socket is cleanly shut down by the peer then reads on this socket
should continue to return zero bytes.  The lwIP socket API does not have
this behaviour (it only returns zero once, then blocks on subsequent calls)
so this patch adds explicit checks and logic for peer closed sockets.

Add --init to the submodule update example, thus, all submodules get
initialised including the nested (--recursive) ones.  Without it there
might not be a submodule init.

Disabling this saves around 6000 bytes of code space and gets the 512k
build fitting in the available flash again (it increased lately due to an
increase in the size of the ESP8266 SDK).

In adcall.py the pyb module may not be imported, so use ADCAll directly.

In dac.py the DAC object now prints more info, so update .exp file.

In spi.py the SPI should be deinitialised upon exit, so the test can run a
second time correctly.

For a given IRQn (eg UART) there's no need to carry around both a PRI and
SUBPRI value (eg IRQ_PRI_UART, IRQ_SUBPRI_UART).  Instead, the IRQ_PRI_UART
value has been changed in this patch to be the encoded hardware value,
using NVIC_EncodePriority.  This way the NVIC_SetPriority function can be
used directly, instead of going through HAL_NVIC_SetPriority which must do
extra processing to encode the PRI+SUBPRI.

For a priority grouping of 4 (4 bits for preempt priority, 0 bits for the
sub-priority), which is used in the stm32 port, the IRQ_PRI_xxx constants
remain unchanged in their value.

This patch also "fixes" the use of raise_irq_pri() which should be passed
the encoded value (but as mentioned above the unencoded value is the same
as the encoded value for priority grouping 4, so there was no bug from this
error).

The problem is the existing code which tries to optimise the
reinitialisation of the DMA breaks the abstraction of the HAL.  For the
STM32L4 the HAL's DMA setup code maintains two private vars (ChannelIndex,
DmaBaseAddress) and updates a hardware register (CCR).

In HAL_DMA_Init(), the CCR is updated to set the direction of the DMA.
This is a problem because, when using the SD Card interface, the same DMA
channel is used in both directions, so the direction bit in the CCR must
follow that.

A quick and effective fix for the L4 is to simply call HAL_DMA_DeInit() and
HAL_DMA_Init() every time.