The SPI flash on current Fomu firmware is slow. Circuitpython runs
XIP from SPI flash, and so execution time can also be slow. Ordinarily
this isn't a problem, however certain operations are time-sensitive.
In particular, USB function needs to be handled quickly in order to
prevent the host from re-enumerating the device.
Place several critical TinyUSB structures into RAM, as well as several
hot functions that are frequently called. This reduces execution time
at the expense of system memory, and greatly improves system stability.
Signed-off-by: Sean Cross <sean@xobs.io>
Because CPython 3.8.0 now produces different output:
- basics/parser.py: CPython does not allow '\\\n' as input.
- import/import_override: CPython imports _io.
PEP479 (see https://www.python.org/dev/peps/pep-0479/) prohibited raising
StopIteration from within a generator (it is turned into a RuntimeError).
This behaviour was introduced in Python 3.5 and in 3.7 was made compulsory.
Until uPy implements PEP479, this patch adds .py.exp files for the relevant
tests so they can be run under Python 3.7.
In Python 3.7 the behaviour of repr() of an exception with one argument
changed: it no longer prints a trailing comma in the argument list. See
https://bugs.python.org/issue30399
This patch modifies tests that rely on this behaviour to not rely on it.
And the python34.py test is updated to include a test for this behaviour
with a .exp file.
ports/atmel-samd/boards/bdmicro_vina_m0/board.c
ports/atmel-samd/boards/bdmicro_vina_m0/mpconfigboard.h
ports/atmel-samd/boards/bdmicro_vina_m0/mpconfigboard.mk
ports/atmel-samd/boards/bdmicro_vina_m0/board.c
This board includes the newly added MX25L51245G 64MB SPI flash chip
for the CircuitPython file system. Also included is the PCF2129T/2,518
high accuracy, battery backed, RTC. Also includes an independent USB
HOST connector in addition to USB DEVICE making it easy to connect USB
peripherals. Designed mainly as a general purpose module to piggy back
onto application-specific boards, it can also run low power standalone
in harsh environments for data logging and other control applications
using an appropriate enclosure.
Has been extensively tested with CircuitPython and works well.
It was fixed as 0/0 even though it used to get it from the current
SPI state. This makes it more explicit with kwargs.
Thanks to magpie_lark and kmatocha on the Adafruit Support forum
for finding the issue: https://forums.adafruit.com/viewtopic.php?f=60&t=162515
