Here is a story of a man-made software issue which really had me stumped.
I’ll pose it as a question , just for fun. The answer to follow in the near future after a few interested readers have had a go, although I’m guessing a number of you may spot this quite easily.
The symptom:
My 6502 SBC has a flashing led routine for testing. It would flash a few leds at a rate of about 1 flash per second, well sometimes. Now and again, it would flash at high speed. I’m using a 1Mhz oscillator.
This is the delay routine:

It is a nested loop routine.
So why did it sometimes produce a correct delay and sometimes a minimal delay?
But wait, there's more:
I also had an 8251 serial comms chip which worked perfectly in another bit of code. Then a few days later it decided not to work.
The part of the code that was blocking things was isolated to this section of code:

It reads a status bit (bit 0) and if the bit is '0', then go back and keep looping until it goes high.
The two problems are related - have you spotted it?
regards
Russell

Vs lbh'q jevggra naq qbyyne bu bar V'q fnl gur ceboyrz va obgu pnfrf jnf zvffvat gur unfu sbe na vzzrqvngr.

(rot13.com)

Ed already pointed out that you're loading your delay registers from a Page Zero location.... but there's nothing that defines if that is supposed to give you a variable delay for multiple uses or just a mistake in coding up the routine. As there's no reference to the Page Zero location, it's likely a mistake.

The second issue is the second routine. If nothing happens with the serial chip, then there's nothing to break it out of the waiting loop, unless you have some ISR that's performing some other task perhaps. The code looks like part of a polled I/O routine, so that can hang the system for a variable length of time.

_________________
Regards, KM
https://github.com/floobydust

Nabgure guvat juvpu pna or n ceboyrz vf vs na vagreehcg freivpr ebhgvar snvyf gb fnir naq erfgber ertvfgref, be punatrf fbzr zrzbel ybpngvba vg fubhyqa'g, be vs vg fgnlf va n ybbc sbe gbb ybat naq hfrf hc gbb zhpu gvzr.

Jurer qvq lbh znc gur 8251 vagb zrzbel? Vg jbhyqa'g unccra gb or pbyyvqvat jvgu $SS, jbhyq vg?

(rot13.com)

Er gur qrynl ebhgvar, V'z pbaivaprq gur nhgube vagraqrq gb hfr na *vzzrqvngr* bcrenaq bs $SS. Vafgrnq, gur pbqr ernqf jung'f ng zrzbel ybpngvba $00SS, naq gur erfhygf ner reengvp orpnhfr guvf ybpngvba unfa'g orra vavgvnyvmrq.

(Ng guvf cbvag V ernq bguref' erfcbafrf, abgvat gung bguref unir vaqrcraqragyl ernpurq gur fnzr pbapyhfvba nobhg gur qrynl ceboyrz. Ohg, er gur frpbaq ceboyrz, fb sne ab-bar unf cebcbfrq gur sbyybjvat. enfpnyfnvybe unf gbyq hf gur gjb ceboyrzf ner eryngrq, fb urer'f zl gurbel.)

PBASVT8251 vf pyrneyl vagraqrq gb ercerfrag na VB ybpngvba, ohg V fhfcrpg n glcb va vgf qrsvavgvba pnhfrf vg, gbb, gb ersre gb na havavgvnyvmrq zrzbel ybpngvba. Guvf zrnaf gur 8251 fgnghf arire rire qbrf trg cbyyrq gb frr vs vg'f ernql, rira ba qnlf jura gur pbqr frrzf gb or jbexvat! V fhfcrpg cbyyvat vf npghnyyl ha-arprffnel, orpnhfr -- sbe jungrire ernfba -- pvephzfgnaprf ner fhpu gung gur 8251 vf NYJNLF ernql ol gur gvzr gur jbhyq-or cbyyvat bpphef!

-- Jeff

Edit: The error in part one has happened before. (Spoiler alert)

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

Hi - Right you are, those elusive '#' symbols had me stumped.
I'm used to Z80 assembler and the use of parenthesis for indirect memory addressing.
Don't forget the hashes!!

Humph. I assumed that $FF contained a constant initialized before entering the routine, to vary the flash rate. The code behaves exactly as written -- it's not fair to test us against your ability to deduce your original intention.

I would also point out that the two pieces of code are not related...

_________________
Regards, KM
https://github.com/floobydust

Maybe, Russell, you're saying that the second loop was originally unreliable because it too lacked the hash?

Sorry , I hadn't replied - Yes the hash omitted caught me out.

Is it just me? Or is half of this thread in Klingon?

Ok, I understand now. Next question is ‘why?’

I used rot13 for the traditional reason, of avoiding spoilers. I like to have a guess, but whether or not I'm right, I don't want to take away from anyone else the pleasure of having their own guess.

Ah! Got ya! Makes sense now.