This patch overhauls the network driver interface. A generic NIC must
provide a set of C-level functions to implement low-level socket control
(eg socket, bind, connect, send, recv). Doing this, the network and
usocket modules can then use such a NIC to implement proper socket
control at the Python level.
This patch also updates the CC3K and WIZNET5K drivers to conform to the
new interface, and fixes some bugs in the drivers. They now work
reasonably well.
Behaviour of array initialisation is subtly different for bytes,
bytearray and array.array when argument has buffer protocol. This patch
gets us CPython conformant (except we allow initialisation of
array.array by buffer with length not a multiple of typecode).
This makes pyboard.py much more useful for long running scripts. When
running a script via pyboard.py, it now waits until the script finishes,
with no timeout. CTRL-C can be used to break out of the waiting if
needed.
pyb.delay and pyb.udelay now use systick if IRQs are enabled, otherwise
they use a busy loop. Thus they work correctly when IRQs are disabled.
The busy loop is computed from the current CPU frequency, so works no
matter the CPU frequency.
By using the buffer protocol for these array operations, we now allow
addition of memoryview objects, and objects with "incompatible"
typecodes (in this case it just adds bytes naively). This is an
extension to CPython which seems sensible. It also reduces the code
size.
Some rodata items can go in iram/irom segment, but not others. With
this patch ESP now has 24256 bytes of heap ram. It passes 228 out of
248 tests from tests/basics directory.
Before, __repl_print__() used libc printf(), while print() used uPy streams
and own printf() implementation. This led to subtle, but confusing
differences in output when just doing "foo" vs "print(foo)" on interactive
prompt.
The reason for having this delay is to reduce power consumption at the
REPL (HAL_Delay calls __WFI to idle the CPU). But stdin_rx_chr has a
__WFI in it anyway, so this delay call is not needed.
By removing this call, the readline input can consume characters much
more quickly (before was limited to 1000 chrs/s), and has much reduced
dependency on the specific port.
