Commit Graph

9537 Commits

Author SHA1 Message Date
Damien George fd10a11c6b py/asmxtensa: Fix bug with order of regs in addi encoding. 2018-08-17 14:11:37 +10:00
Damien George f774614110 tests/micropython: Add tests for try and with blocks under native/viper. 2018-08-17 14:11:36 +10:00
Damien George a3de776486 py/emitnative: Optimise and improve exception handling in native code.
Prior to this patch, native code would use a full nlr_buf_t for each
exception handler (try-except, try-finally, with).  For nested exception
handlers this would use a lot of C stack and be rather inefficient.

This patch changes how exceptions are handled in native code by setting up
only a single nlr_buf_t context for the entire function, and then manages a
state machine (using the PC) to work out which exception handler to run
when an exception is raised by an nlr_jump.  This keeps the C stack usage
at a constant level regardless of the depth of Python exception blocks.

The patch also fixes an existing bug when local variables are written to
within an exception handler, then their value was incorrectly restored if
an exception was raised (since the nlr_jump would restore register values,
back to the point of the nlr_push).

And it also gets nested try-finally+with working with the viper emitter.

Broadly speaking, efficiency of executing native code that doesn't use
any exception blocks is unchanged, and emitted code size is only slightly
increased for such function.  C stack usage of all native functions is
either equal or less than before.  Emitted code size for native functions
that use exception blocks is increased by roughly 10% (due in part to
fixing of above-mentioned bugs).

But, most importantly, this patch allows to implement more Python features
in native code, like unwind jumps and yielding from within nested exception
blocks.
2018-08-16 13:56:36 +10:00
Damien George 2964b41c28 py/asm*: Support assembling code to jump to a register, and get PC+off.
Useful for position independent code, and implementing state machines.
2018-08-16 13:45:24 +10:00
Damien George f7d6108d1a py/asmxtensa: Handle function entry/exit when stack use larger than 127. 2018-08-16 13:43:36 +10:00
Damien George 8c49995398 py/emitnative: Use small tables to simplify handling of local regs. 2018-08-15 10:55:11 +10:00
Damien George 056e0b6293 stm32/spi: Add implementation of low-level SPI protocol.
Can be used, for example, to configure external SPI flash using a hardware
SPI interface (code to be put in a board's bdev.c file):

    STATIC const spi_proto_cfg_t hard_spi_bus = {
        .spi = &spi_obj[5],
        .baudrate = 10000000,
        .polarity = 0,
        .phase = 0,
        .bits = 8,
        .firstbit = SPI_FIRSTBIT_MSB,
    };

    STATIC mp_spiflash_cache_t spi_bdev_cache;

    const mp_spiflash_config_t spiflash_config = {
        .bus_kind = MP_SPIFLASH_BUS_SPI,
        .bus.u_spi.cs = pin_A0,
        .bus.u_spi.data = (void*)&hard_spi_bus,
        .bus.u_spi.proto = &spi_proto,
        .cache = &spi_bdev_cache,
    };

    spi_bdev_t spi_bdev;
2018-08-14 22:10:43 +10:00
Damien George 01ce2e1682 unix/Makefile: Enable ussl module with nanbox build. 2018-08-14 21:53:06 +10:00
Damien George 206c65f22c extmod/modussl_axtls: Use MP_ROM_PTR for objects in allowed args array. 2018-08-14 21:47:07 +10:00
Damien George b8b2525576 extmod/modbtree: Update to work with new mp_stream_posix_XXX signatures. 2018-08-14 17:41:23 +10:00
Damien George 9ab816d676 py/stream: Adjust mp_stream_posix_XXX to take void*, not mp_obj_t.
These POSIX wrappers are assumed to be passed a concrete stream object so
it is more efficient (eg on nan-boxing builds) to pass in the pointer
rather than mp_obj_t, because then the users of these functions only need
to store a void* (and mp_obj_t may be wider than a pointer).  And things
would be further improved if the stream protocol functions eventually took
a pointer as their first argument (instead of an mp_obj_t).

This patch is a step to getting ussl/axtls compiling on nan-boxing builds.

See issue #3085.
2018-08-14 17:36:08 +10:00
Paul Sokolovsky ab78fe0eb9 mpy-cross/Makefile: Also undefine MICROPY_FORCE_32BIT and CROSS_COMPILE.
mpy-cross is a host, not target binary. It should not be build with the
target compiler, compiler options and other settings. For example,

If someone currently tries to build from pristine checkout the unix port
with the following command:

    make CROSS_COMPILE=arm-linux-gnueabihf-

then mpy-cross will be built with arm-linux-gnueabihf-gcc and of course
won't run on the host, leading to overall build failure.

This situation was worked around for some options in 1d8c3f4cff, so add
MICROPY_FORCE_32BIT and CROSS_COMPILE to that set too.
2018-08-14 17:20:18 +10:00
Damien George 8300be6d0f stm32/spi: Split out pyb.SPI and machine.SPI bindings to their own files
The aim here is to have spi.c contain the low-level SPI driver which is
independent (not fully but close) of MicroPython objects and methods, and
the higher-level bindings are separated out to pyb_spi.c and machine_spi.c.
2018-08-14 17:11:07 +10:00
Damien George 48d736f491 esp32: Update to latest ESP IDF.
Among other things, this requires putting bootloader object files in to
their relevant .a archive, so that they can be correctly referenced by the
ESP IDF's linker script.
2018-08-14 16:45:37 +10:00
Damien George a785a3dbfb py/objarray: Allow to build again when bytearray is disabled. 2018-08-14 16:23:21 +10:00
Damien George 91041945c9 py/gc: In gc_alloc, reset n_free var right before search for free mem.
Otherwise there is the possibility that n_free starts out non-zero from the
previous iteration, which may have found a few (but not enough) free blocks
at the end of the heap.  If this is the case, and if the very first blocks
that are scanned the second time around (starting at
gc_last_free_atb_index) are found to give enough memory (including the
blocks at the end of the heap from the previous iteration that left n_free
non-zero) then memory will be allocated starting before the location that
gc_last_free_atb_index points to, most likely leading to corruption.

This serious bug did not manifest itself in the past because a gc_collect
always resets gc_last_free_atb_index to point to the start of the GC heap,
and the first block there is almost always allocated to a long-lived
object (eg entries from sys.path, or mounted filesystem objects), which
means that n_free would be reset at the start of the search loop.

But with threading enabled with the GIL disabled it is possible to trigger
the bug via the following sequence of events:

1. Thread A runs gc_alloc, fails to find enough memory, and has a non-zero
   n_free at the end of the search.
2. Thread A calls gc_collect and frees a bunch of blocks on the GC heap.
3. Just after gc_collect finishes in thread A, thread B takes gc_mutex and
   does an allocation, moving gc_last_free_atb_index to point to the
   interior of the heap, to a place where there is most likely a run of
   available blocks.
4. Thread A regains gc_mutex and does its second search for free memory,
   starting with a non-zero n_free.  Since it's likely that the first block
   it searches is available it will allocate memory which overlaps with the
   memory before gc_last_free_atb_index.
2018-08-14 16:11:21 +10:00
forester3 02fbb0a455 stm32/boards/STM32F7DISC: Enable onboard SDRAM.
The default SYSCLK frequency is reduced to 192MHz because SDRAM requires it
to be 200MHz or less.
2018-08-14 16:04:10 +10:00
forester3 502c410214 stm32/boards/STM32F429DISC: Add burst len and autorefresh to SDRAM cfg.
To align with recent changes to sdram.c.
2018-08-14 16:03:13 +10:00
forester3 e562f99263 stm32/sdram: Allow additional config by a board, and tune MPU settings.
- Allow configuration by a board of autorefresh number and burst length.
- Increase MPU region size to 8MiB.
- Make SDRAM region cacheable and executable.
2018-08-14 16:00:14 +10:00
Damien George b18fa1e606 docs/library/machine.UART.rst: Specify optional txbuf and rxbuf args.
If a port would like to expose the configuration of transmit and/or receive
buffers then it can use these arguments.
2018-08-14 15:21:54 +10:00
Paul Sokolovsky fe1ef507ef unix/Makefile: coverage: Explicitly build "axtls" too.
"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.
2018-08-14 15:10:52 +10:00
Paul Sokolovsky bb28fe7b7b py/py.mk: Don't hardcode path to libaxtls.a.
Use -L$(BUILD), not -Lbuild. Otherwise, builds for different archs/subarchs
using different values of BUILD may fail.
2018-08-14 15:10:52 +10:00
stijn 3f9d3e120b windows/msvc: Support custom compiler for header generation.
Use overrideable properties instead of hardcoding the use of the
default cl executable used by msvc toolsets. This allows using
arbitrary compiler commands for qstr header generation.
The CLToolExe and CLToolPath properties are used because they are,
even though absent from any official documentation, the de-facto
standard as used by the msvc toolsets themselves.
2018-08-14 15:07:19 +10:00
Damien George cbec17f2cd py/compile: For dynamic compiler, widen literal 1 to get correct shift.
Without this patch, on 64-bit architectures the "1 << (small_int_bits - 1)"
is computed using only 32-bit values (since small_int_bits is a uint8_t)
and so will overflow (and give the wrong result) if small_int_bits is
larger than 32.
2018-08-13 23:34:47 +10:00
Damien George 86e0b25532 stm32/spi: Round up prescaler calc to never exceed requested baudrate.
Requesting a baudrate of X should never configure the peripheral to have a
baudrate greater than X because connected hardware may not be able to
handle higher speeds.  This patch makes sure to round the prescaler up so
that the actual baudrate is rounded down.
2018-08-10 16:39:47 +10:00
stijn ca0d78cebb run-tests: Make .exp and .out file names unique by prefixing with dir.
Input files like basics/string_format.py and float/string_format.py have
the same basename so using that name for writing the output (.exp and .out
files) when both tests fail, results in the output of the first one being
overwritten.

Avoid this by using unique names for the output, replacing path characters
with underscores.
2018-08-10 16:33:42 +10:00
David Lechner 3fccd78aca stm32/dma: Fix spelling of "corresponding" in two locations. 2018-08-10 16:26:25 +10:00
Martin Dybdal 5ed8226e02 tools/pyboard.py: Change base class of PyboardError to Exception.
Following standard practice for defining custom exceptions.
2018-08-10 16:23:38 +10:00
roland c1c798fbc3 drivers/cc3000: Use cc3000_time_t instead of time_t for custom typedef.
Otherwise it can clash with time_t from the C standard include headers.
2018-08-08 16:37:26 +10:00
Damien George 17b512020b py/emitnative: Allocate space for local stack info as it's needed. 2018-08-07 16:19:38 +10:00
Damien George 652a58698e py/emitnative: Simplify handling of exception objects from nlr_buf_t.
There is no need to have three copies of the exception object on the top of
the native value stack.  Instead, the values on the stack should be the
first two items in an nlr_buf_t: the prev pointer and the ret_val pointer.
This is all that is needed and is what the rest of the native emitter
expects is on the stack.

This patch is essentially an optimisation.  Behaviour is unchanged,
although the stack layout for native exception handling now makes more
sense.
2018-08-06 14:44:33 +10:00
Damien George 3bef7bd782 py/emitnative: Fix native locals stack to start at correct location.
A native function allocates space on its C stack for mp_code_state_t,
followed by its Python stack, then its locals.  This patch makes sure that
the native function actually starts at the start of its Python stack,
rather than at the start of mp_code_state_t (which didn't lead to any
issues so far because the mp_code_state_t is unused after the native
function sets itself up).
2018-08-04 22:41:35 +10:00
Damien George 1c0bd46d1d py/asmx86: Use generic emit function to simplify cmp emit function. 2018-08-04 22:26:14 +10:00
Damien George ce786da196 tests/run-tests: Enable bool1.py test with native emitter.
It should work reliably now.
2018-08-04 22:19:04 +10:00
Damien George 49529f22d4 tests/micropython/viper_cond: Add test for large int as bool. 2018-08-04 22:16:24 +10:00
Damien George 10830059c5 py/emitnative: Fix x86 native zero checks by comparing full word.
On x86 archs (both 32 and 64 bit) a bool return value only sets the 8-bit
al register, and the higher bits of the ax register have an undefined
value.  When testing the return value of such cases it is required to just
test al for zero/non-zero.  On the other hand, checking for truth or
zero/non-zero on an integer return value requires checking all bits of the
register.  These two cases must be distinguished and handled correctly in
generated native code.  This patch makes sure of this.

For other supported native archs (ARM, Thumb2, Xtensa) there is no such
distinction and this patch does not change anything for them.
2018-08-04 22:03:49 +10:00
Damien George 4b1e8bdebd py/emitnative: Factor common code for native jump helper. 2018-08-04 21:45:24 +10:00
Peter Hinch 163bacd1e8 docs/library/machine.I2C.rst: Clarify availability of primitive I2C ops. 2018-08-04 15:53:12 +10:00
Ayke van Laethem 0d7a088039 tools/pyboard: Run exec: command as a string.
The Python documentation recommends to pass the command as a string when
using Popen(..., shell=True).  This is because "sh -c <string>" is used to
execute the command and additional arguments after the command string are
passed to the shell itself (not the executing command).

https://docs.python.org/3.5/library/subprocess.html#subprocess.Popen
2018-08-04 15:45:23 +10:00
Ayke van Laethem 6572029dc0 tests: Make tests work on targets without float support. 2018-08-04 15:14:23 +10:00
Damien George 7be5bb3672 stm32/adc: Fix ADC reading on F0 MCUs to only sample a single channel.
And increase sampling time to get better results for internal channels.
2018-08-04 13:33:02 +10:00
Damien George c62b23094f stm32/adc: Disable VBAT in read channel helper function.
Prior to this patch, if VBAT was read via ADC.read() or
ADCAll.read_channel(), then it would remain enabled and subsequent reads
of TEMPSENSOR or VREFINT would not work.  This patch makes sure that VBAT
is disabled for all cases that it could be read.
2018-08-04 13:25:43 +10:00
Ayke van Laethem e755bd4932
nrf/uart: Fix UART.writechar() to write just 1 byte. 2018-08-02 22:21:24 +02:00
Ayke van Laethem 60a05485cb
nrf/uart: Remove unused UART.char_width field.
Also, clean up some code.
Code size change:
nrf51: -24
nrf52: -28
2018-08-02 22:18:16 +02:00
Stig Bjørlykke 2cf2ad943e nrf: Use separate config for each PWM instance.
The hard_configs table has entries for each PWM instance. Use them.
2018-08-02 20:36:18 +02:00
Damien George b630dfcc1d py: Fix compiling with debug enabled and make more use of DEBUG_printf.
DEBUG_printf and MICROPY_DEBUG_PRINTER is now used instead of normal
printf, and a fault is fixed in mp_obj_class_lookup with debugging enabled;
see issue #3999.  Debugging can now be enabled on all ports including when
nan-boxing is used.
2018-08-02 14:17:24 +10:00
Damien George da2d2b6d88 py/mpconfig.h: Introduce MICROPY_DEBUG_PRINTER for debugging output.
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.
2018-08-02 14:04:44 +10:00
Stig Bjørlykke 0c161691b4 nrf: Correct index checking of ADC/PWM/RTCounter instances.
Avoid trying to use ADC, PWM and RTCounter instances which is
one past last available, because this will give a HardFault.
2018-08-01 14:56:07 +02:00
Stig Bjørlykke 7f0c5f2ef9 nrf: Enable all PWM, RTC and Timer instances for nrf52840.
The NRF52 define only covers nrf52832, so update the define checks
to use NRF52_SERIES to cover both nrf52832 and nrf52840.

Fixed machine_hard_pwm_instances table in modules/machine/pwm.c

This enables PWM(0) to PWM(3), RTCounter(2), Timer(3) and Timer(4),
in addition to NFC reset cause, on nrf52840.
2018-08-01 14:36:07 +02:00
Stig Bjørlykke b6e49da407
nrf/uos: Add mbfs __enter__ and __exit__ handlers.
This will make 'with open('file', 'r') as f:' work by properly close
the file after the suite is finished.
2018-08-01 14:12:34 +02:00