Loop in code/tight leads to hard fault

[mterber]

Hi,

I compiled the following Céu code for an STM32F103 ARM Cortex-M3 microcontroller using the Céu compiler "ceu 0.30 (8540111)":

native/pre do
    ##include "diag/Trace.h"
end

native/nohold _trace_puts;
native/nohold _trace_printf;

code/tight Test (var u32 a) -> none do
	loop do
		if a < 1 then break; end
		a = a - 1;
		_trace_puts("bar");
	end
end

loop do
	_trace_puts("foo");
	await 1s;
	call Test(3);
end

When I run the code on the microcontroller, I get the following output:

foo
bar
[HardFault]
Stack frame:
 R0 =  00000001
 R1 =  08004EC0
 R2 =  00000000
 R3 =  00000002
 R12 = 20004E50
 LR =  08001433
 PC =  080034B8
 PSR = 21000000
FSR/FAR:
 CFSR =  00000400
 HFSR =  40000000
 DFSR =  0000000A
 AFSR =  00000000
Misc
 LR/EXC_RETURN= FFFFFFF9

So as soon as the second iteration of the loop in the code/tight is executed a hard fault error makes the controller to halt. This behavior is reproducible and happens on every run. If I remove the loop and just output "bar" the error is gone.

[fsantanna]

Thank you for the report.
I can confirm it is a bug.

This code also reproduces the bug:

code/tight Test (var int a) -> none do
    loop do
        if a < 1 then break; end
        a = a - 1;
    end
end
call Test(3);
escape 1;

(Surprising that there's no similar test case.)

[fsantanna]

Fixed on master.
It is a little bit ahead of v0.30 but it should not be a problem.

[ceremcem]

@mterber I want to create a test app with Ceu on STM32 (I have STM32F103 and F030 eval boards). Do you have a test repo for a quick startup?