[PIC24]REPL error of assertion"emit->stack_zise == 0"

[QuanLin]

Hi,

I'm trying to port MicroPython to PIC24 (PIC24FJ1024GB606), a 16-bit micro from Microchip.
I'm using MicroPython version 1.9.2, the same version used for MicroBit.
Compiler: XC16 v1.35

The experiment project compiles and run the REPL on UART.
Here's some brief result:

Very simple math works like:
>>> 1+1
2
>>> 24*34
816

Others fails:
>>> 256*256
assertion "emit->stack_size == 0" failed: file "source/py/emitbc.c", line 406, function: mp_emit_bc_end_pass
>>> a=1
assertion "exc_sp == exc_stack - 1" failed: file "source/py/vm.c", line 1095, function: mp_execute_bytecode
>>> dir()
assertion "tok == MP_TOKEN_ELLIPSIS" failed: file "source/py/emitbc.c", line 525, function: mp_emit_bc_load_const_tok

Can anyone kindly tell me what I'm missing?

[jimmo]

Hi,

I'm trying to port MicroPython to PIC24 (PIC24FJ1024GB606), a 16-bit micro from Microchip.

Are you using the existing pic16bit port in the main codebase? i.e. https://github.com/micropython/micropyt ... s/pic16bit

I'm using MicroPython version 1.9.2, the same version used for MicroBit.

Any particular reason to use 1.9.2 rather than the latest release (1.11) or master?

Others fails:
>>> 256*256

The difference between the two is that the first example works entirely using small ints (i.e. the inline representation in mp_obj_t) where the second needs to allocate an object. I'm guessing that something's going wrong leading to something like stack corruption.

[QuanLin]

Hi Jimmo,

Thank you for the reply.
This afternoon I just tried the latest version (1.11.0) from https://github.com/micropython/micropyt ... s/pic16bit
It compiled as well and gave me the same result. The REPL works on UART and small number math works.
The errors for the others are the same.

I did not use the Makefile from here(https://github.com/micropython/micropyt ... t/Makefile).
Instead I used MPLAB X IDE and created a new project and it generated its own Makefile
I did take care of all of the compiler (XC16 v1.35) flags like: -std=gnu99 -mlarge-code -nostdlib, otherwise it won't compile.

In the project I included the whole "py" folder and
a few necessary "lib" ("mp-readline" and some "utils") and
a few necessary extmod(virtpin.h and virtpin.c, utime_mphal.h and utime_mphal.c)

It compiles with a few warnings and the generated hex file works.

I uploaded the zipped project files to here: https://github.com/jacklinquan/PIC24FJ1024GB606_test2
To open it, MPLAB X IDE v5.10 or above is needed. XC16 v1.35 is needed.

[QuanLin]

@jimmo
@Damien

I've confirmed my findings with the micropython port for pic16 from https://github.com/micropython/micropyt ... s/pic16bit

I built this port with PIC24FJ1024GB606 and the same Makefile(the path of XC16 changed).
You can find it here: https://github.com/jacklinquan/micropyt ... j1024gb606

Here's some brief test result:
When NDEBUG is defined in the Makefile:
##################
>>> 1+1
2
>>> a=1
>>> a
1
>>> b=2
>>> c=a+b
>>> c
Traceback (most recent call last):
File "<stdin>", in <module>
NameError: name not defined
>>> 256*256
Traceback (most recent call last):
File "<stdin>", in <module>
TypeError: unsupported type for operator
>>> 25*25
625
##################

When NDEBUG is NOT defined in the Makefile:
##################
>>> 1+1
2
>>> a=1
assertion "exc_sp == exc_stack - 1" failed: file "../../py/vm.c", line 1143, function: mp_execute_bytecode
MicroPython v1.11 on 2019-10-03; EXPERIMENT with PIC24FJ1024GB606 (Note: It rebooted here.)
>>> 25*25
625
>>> 256*256
assertion "emit->stack_size == 0" failed: file "../../py/emitbc.c", line 426, function: mp_emit_bc_end_pass
MicroPython v1.11 on 2019-10-03; EXPERIMENT with PIC24FJ1024GB606 (Note: It rebooted here.)
>>> dir
<function>
>>> dir()
assertion "emit->stack_size == 0" failed: file "../../py/emitbc.c", line 426, function: mp_emit_bc_end_pass
MicroPython v1.11 on 2019-10-03; EXPERIMENT with PIC24FJ1024GB606 (Note: It rebooted here.)
>>> int
<class 'int'>
>>> int('1')
assertion "emit->stack_size == 0" failed: file "../../py/emitbc.c", line 426, function: mp_emit_bc_end_pass
MicroPython v1.11 on 2019-10-03; EXPERIMENT with PIC24FJ1024GB606 (Note: It rebooted here.)
>>> s='hello'
assertion "exc_sp == exc_stack - 1" failed: file "../../py/vm.c", line 1143, function: mp_execute_bytecode
MicroPython v1.11 on 2019-10-03; EXPERIMENT with PIC24FJ1024GB606 (Note: It rebooted here.)
>>> l=[1,2,3]
assertion "emit->stack_size == 0" failed: file "../../py/emitbc.c", line 426, function: mp_emit_bc_end_pass
MicroPython v1.11 on 2019-10-03; EXPERIMENT with PIC24FJ1024GB606 (Note: It rebooted here.)
##################

For the moment I just have PIC24FJ1024GB606 at hand. Could you confirm my findings with your boards?
It looks like the port for pic16bit is broken.
I'm opening an issue on github.

Thanks!