When going out of memory-mapped mode to do a control transfer to the QSPI
flash, the MPU settings must be changed to forbid access to the memory
mapped region. And any ongoing transfer (eg memory mapped continuous read)
must be aborted.
The Cortex-M7 CPU will do speculative loads from any memory location that
is not explicitly forbidden. This includes the QSPI memory-mapped region
starting at 0x90000000 and with size 256MiB. Speculative loads to this
QSPI region may 1) interfere with the QSPI peripheral registers (eg the
address register) if the QSPI is not in memory-mapped mode; 2) attempt to
access data outside the configured size of the QSPI flash when it is in
memory-mapped mode. Both of these scenarios will lead to issues with the
QSPI peripheral (eg Cortex bus lock up in scenario 2).
To prevent such speculative loads from interfering with the peripheral the
MPU is configured in this commit to restrict access to the QSPI mapped
region: when not memory mapped the entire region is forbidden; when memory
mapped only accesses to the valid flash size are permitted.
This fixes compiling for older architectures (e.g. armv5tej).
According to [1], the limit of R0-R7 for the STR and LDR instructions is
tied to the Thumb instruction set and not any specific processor
architectures.
[1]: http://www.keil.com/support/man/docs/armasm/armasm_dom1361289906890.htm
When compiled with hard float the system should enable FP access when it
starts or else FP instructions lead to a fault. But this minimal port does
not enable (or use) FP and so, to keep it minimal, switch to use soft
floating point. (This became an issue due to the recent commit
34c04d2319 which saves/restores FP registers
in the NLR state.)
misc_aes.py and misc_mandel.py are adapted from sources in this repository.
misc_pystone.py is the standard Python pystone test. misc_raytrace.py is
written from scratch.
This benchmarking test suite is intended to be run on any MicroPython
target. As such all tests are parameterised with N and M: N is the
approximate CPU frequency (in MHz) of the target and M is the approximate
amount of heap memory (in kbytes) available on the target. When running
the benchmark suite these parameters must be specified and then each test
is tuned to run on that target in a reasonable time (<1 second).
The test scripts are not standalone: they require adding some extra code at
the end to run the test with the appropriate parameters. This is done
automatically by the run-perfbench.py script, in such a way that imports
are minimised (so the tests can be run on targets without filesystem
support).
To interface with the benchmarking framework, each test provides a
bm_params dict and a bm_setup function, with the later taking a set of
parameters (chosen based on N, M) and returning a pair of functions, one to
run the test and one to get the results.
When running the test the number of microseconds taken by the test are
recorded. Then this is converted into a benchmark score by inverting it
(so higher number is faster) and normalising it with an appropriate factor
(based roughly on the amount of work done by the test, eg number of
iterations).
Test outputs are also compared against a "truth" value, computed by running
the test with CPython. This provides a basic way of making sure the test
actually ran correctly.
Each test is run multiple times and the results averaged and standard
deviation computed. This is output as a summary of the test.
To make comparisons of performance across different runs the
run-perfbench.py script also includes a diff mode that reads in the output
of two previous runs and computes the difference in performance. Reports
are given as a percentage change in performance with a combined standard
deviation to give an indication if the noise in the benchmarking is less
than the thing that is being measured.
Example invocations for PC, pyboard and esp8266 targets respectively:
$ ./run-perfbench.py 1000 1000
$ ./run-perfbench.py --pyboard 100 100
$ ./run-perfbench.py --pyboard --device /dev/ttyUSB0 50 25
With both MICROPY_PERSISTENT_CODE_SAVE and MICROPY_PERSISTENT_CODE_LOAD
enabled the code fails to compile, due to undeclared 'n_obj'. If
MICROPY_EMIT_NATIVE is disabled there are more errors due to the use of
undefined fields in mp_raw_code_t.
This patch fixes such compilation by avoiding undefined fields.
MICROPY_EMIT_NATIVE was changed to MICROPY_EMIT_MACHINE_CODE in this file
to match the mp_raw_code_t definition.
Change static LED functions to lowercase names, and trim down source code
lines for variants of MICROPY_HW_LED_COUNT. Also rename configuration for
MICROPY_HW_LEDx_LEVEL to MICROPY_HW_LEDx_PULLUP to align with global PULLUP
configuration.
Commit 9e68eec8ea introduced a regression
where the PID of the USB device would be 0xffff if the default value was
used. This commit fixes that by using a signed int type.
This saves time when building on Travis CI: unconditionally fetching all
submodules takes about 40 seconds, but not all are needed for any given
port, so only fetch as necessary.
Entering a bootloader (ST system bootloader, or custom mboot) from software
by directly branching to it is not reliable, and the reliability of it
working can depend on the peripherals that were enabled by the application
code. It's also not possible to branch to a bootloader if the WDT is
enabled (unless the bootloader has specific provisions to feed the WDT).
This patch changes the way a bootloader is entered from software by first
doing a complete system reset, then branching to the desired bootloader
early on in the start-up process. The top two words of RAM (of the stack)
are reserved to store flags indicating that the bootloader should be
entered after a reset.
WIFI_REASON_AUTH_FAIL does not necessarily mean the password is wrong, and
a wrong password may not lead to a WIFI_REASON_AUTH_FAIL error code. So to
improve reliability connecting to a WLAN always reconnect regardless of the
error.
These s16-s21 registers are used by gcc so need to be saved. Future
versions of gcc (beyond v9.1.0), or other compilers, may eventually need
additional registers saved/restored.
See issue #4844.
This updates ESP IDF to use v3.3-beta3. And also adjusts README.md to
point to stable docs which provide a link to download the correct toolchain
for this IDF version, namely 1.22.0-80-g6c4433a-5.2.0
Previously the end of the heap was the start (lowest address) of the stack.
With the changes in this commit these addresses are now independent,
allowing a board to place the heap and stack in separate locations.
Damien George