not want to try to modify for insert some custom instructions? i think that this assembler have already some custom functions!
i see that some users of this thread intervent in this thread but none did also me to intervent with asm!
Sorry, I don't understand. STS has 32-bit Opcode. No matter what, the byte at the address 0x20 stays 0x00. 0x20 is the first byte in 64 I/O register space and it is reserved.
Of course, as I wrote, it was in Atmel Studio, not real chip.
Yes, it was simulation. I don't have anything usable with 328P right now.
Almel Studio is pretty accurate.
You try it. What's the worst that can happen?
??
because really i ask myself because "reserved" !
YESSSSSS!!!!!
This is atmega2560 https://ww1.microchip.com/downloads/en/devicedoc/atmel-2549-8-bit-avr-microcontroller-atmega640-1280-1281-2560-2561_datasheet.pdf
So the ATmega328 follows the same design approach as the CH32V001, where a unified register map reserves spaces for missing GPIO ports on lower pin-count packages.
Also ch32v001 is interesting to study in asm , and write a uploader based on webusb i understand that is some sample , but problem is a assembler in js because have more instructions.
Anyway , i did , first i begin to write by hand a avr assembler (without help of ai) but after days of time spend i find avrlass.js e avrdass.js that was ready to use , so i not continue the tool Create and upload Arduino hex file directly from a web page
This first tool only these instructions accepted (i stop here because i find avrlass.js):
case 'LDI': code.push(1,1);break;
case 'OUT': code.push(1,1);break;
case 'RJMP':code.push(1,1);break;
case 'BRNE':code.push(1,1);break;
case 'DEC':code.push(1,1);break;
case 'INC':code.push(1,1);break;
case 'EOR':code.push(1,1);break;
case 'RET':code.push(1,1);break;
case 'RCALL':code.push(1,1);break;
case 'CBI':code.push(1,1);break;
case 'SBI':code.push(1,1);break;
case 'SBIC':code.push(1,1);break;
case 'SBRC':code.push(1,1);break;
case 'NOP':code.push(1,1);break;
My m328p board returns mostly 0xB8 from PORTA and DDRA, no matter what is written to them....
#define PINA _SFR_IO8(0x00)
#define DDRA _SFR_IO8(0x01)
#define PORTA _SFR_IO8(0x02)
void setup() {
Serial.begin(115200);
Serial.println("Test (non-existant) Port A registers");
for (int i = 0; i < 256; i++) {
DDRA = i;
PORTA = i;
Serial.print("i=0x"); Serial.print(i, HEX);
Serial.print(" DDRA=0x"); Serial.print((unsigned int)DDRA, HEX);
Serial.print(" PORTA=0x"); Serial.println((unsigned int)PORTA, HEX);
delay(500);
}
}
void loop() {}
says:
Test (non-existant) Port A registers
i=0x0 DDRA=0xB8 PORTA=0xB8
i=0x1 DDRA=0xB8 PORTA=0xB8
i=0x2 DDRA=0xB8 PORTA=0x2
i=0x3 DDRA=0xB8 PORTA=0xB8
i=0x4 DDRA=0xB8 PORTA=0xB8
i=0x5 DDRA=0xB8 PORTA=0xB8
i=0x6 DDRA=0xB8 PORTA=0x2
i=0x7 DDRA=0xB8 PORTA=0xB8
i=0x8 DDRA=0xB8 PORTA=0xB8
i=0x9 DDRA=0xB8 PORTA=0xB8
i=0xA DDRA=0xB8 PORTA=0xB8
i=0xB DDRA=0xB8 PORTA=0xB8
i=0xC DDRA=0xB8 PORTA=0xB8
i=0xD DDRA=0xB8 PORTA=0xB8
i=0xE DDRA=0xB8 PORTA=0x2
i=0xF DDRA=0xB8 PORTA=0xB8
i=0x10 DDRA=0xB8 PORTA=0xB8
(all remaining values yield B8)
It'd probably be "interesting" to see if the results are the same on other 28pin AVRs....
I tried your sketch, results similar, irregularities similar.
I modified your sketch and ran it:
#define PINA _SFR_IO8(0x00)
#define DDRA _SFR_IO8(0x01)
#define PORTA _SFR_IO8(0x02)
void setup() {
Serial.begin(115200);
Serial.println("Test (non-existant) Port A registers");
for (int i = 0; i < 256; i++) {
DDRA = i;
PORTA = i;
uint8_t pinA = PINA;
uint8_t ddrA = DDRB;
uint8_t portA = PORTA;
Serial.print("i=0x"); Serial.print(i, HEX);
Serial.print(" PINA=0x"); Serial.print(pinA, HEX);
Serial.print(" DDRA=0x"); Serial.print(ddrA, HEX);
Serial.print(" PORTA=0x"); Serial.println(portA, HEX);
delay(100);
}
}
void loop() {}
22:26:07.342 -> Test (non-existant) Port A registers
22:26:07.342 -> i=0x0 PINA=0xB8 DDRA=0x0 PORTA=0x0
22:26:07.444 -> i=0x1 PINA=0x6 DDRA=0x0 PORTA=0x0
22:26:07.560 -> i=0x2 PINA=0x6 DDRA=0x0 PORTA=0x0
22:26:07.663 -> i=0x3 PINA=0x6 DDRA=0x0 PORTA=0x0
22:26:07.732 -> i=0x4 PINA=0x6 DDRA=0x0 PORTA=0x0
22:26:07.857 -> i=0x5 PINA=0x6 DDRA=0x0 PORTA=0x0
without further irregularities.
I speculate that the databus has pulldown resistors and all pripherials have an open drain output.
The databus lines, connecting all peripherials, will have significant parasitic capacitance.
Suppose the RC-time is just not short enough. The 0xB8 may be the last value on the databus after Serial.print() and other irregularities may be remnants from the millis()interrupt.
Still, I have the feeling there is more to it.
For they that have 100% confident ai:
2. sbi and cbi instructions:
sbi 2,5 → Set Bit in I/O register: sets to 1 bit 5 of the I/O register at address 2.
Address 2 = PORTB (on ATmega328P)
Bit 5 = pin 13 (Arduino built-in LED)
sbi 5,5 → Sets bit 5 of the I/O register at address 5.
Address 5 = PORTD (on ATmega328P)
Bit 5 = pin 5 of Arduino (if used with an external LED)
You still trust AI?
You'd better read the datasheet.
No... I put this for those who say I write Assembly with AI.
Perhaps for C++ it's easier, but for ASM it's a battle with AI. It's not easy to write pure ASM with AI.
Anyway, lately I see AI is more friendly with sbi 5,5. Perhaps also for my posts. I wrote many months of tutorials with sbi 5,5, but a long time ago it was a battle with AI that didn't accept sbi 5,5, but only PORTB and PB5!
Before, AI used to insist that Assembly is useless and too complicated, that it's not even worth trying! But since I created my AVR tool and many tutorials, it seems that AI's mindset is finally changing a bit!
No... i put this for them that did that i use ai for create asm code!
Perhaps for c++ is more easy , but for asm is a battle with ai ,is not easy to write pure asm with ai.
Anyway ,ultimate i see ai more friendly with sbi 5,5 perhaps also for my posts, i write many months tutorial with sbi 5,5 , but more time ago was a battle with ai that not accepted sbi 5,5, but PORTB and PB5 only!
Okay. Now you're just making things up.
@EmilyJane ask ai for "costycnc avr tool"
im curious what did ai!
I
It says " the underlying idea is excellent, but I would use the present version cautiously because some instructional material contains significant technical errors."
Huh!, my impression of AI has changed. It's more intuitive than I gave it credit for.
If it doesn't cite which page, which instruction, which register, and what the error is, the phrase “significant technical errors” is basically an accusation without evidence.
I'm paying the subscription for my AI. Why would I give you advice for free?
Anyway ,if your payed professional AI (CLAUDE?) did only "some" , i can to go to sleep reassured! For a carpenter is more that enough!
Yes, we're well aware of your low standards of excellence. 