Garbled display with Pico RP2040

[caternuson]

Re this thread: https://forums.adafruit.com/viewtopic.php?f=47&t=187543

Issue recreated locally. Getting identical wrong results, so not noise. Works fine on Itsy M4, so not 3v vs 5v. Most likely something RP2040 specific.

[ladyada]

!? thats quite wierd

[caternuson]

Looks potentially related to use of pgm_read_word. Uncommenting these debug prints: https://github.com/adafruit/Adafruit_LED_Backpack/blob/508c0e6694d5197e598f81b5ca0337f58ddd1940/Adafruit_LEDBackpack.cpp#L338-L342 Get the following:

Itsy M4 (expected output) 65 / 'A' = 0xF7 66 / 'B' = 0x128F 67 / 'C' = 0x39 68 / 'D' = 0x120F

RP2040 (garbled output) 65 / 'A' = 0x600 66 / 'B' = 0x6 67 / 'C' = 0x2000 68 / 'D' = 0x220

I've scoped this and the I2C traffic looks fine. It's clocking out exactly what's in the buffer. So it's more an internal data type/size thing specific to RP2040.

[ladyada]

sounds like pgmreadword doesnt do what it should - which is nothing?

[caternuson]

Opened issue with core: https://github.com/earlephilhower/arduino-pico/issues/447

[earlephilhower]

Release 1.9.15 or arduino-pico should fix this. The pgm_read macros support misaligned access (since the RP2040 HW does not) and having base + offset exposed some missing parens().

[caternuson]

@earlephilhower Thanks for the quick fix! This resolved the forum issue. :)

[ladyada]

@caternuson now you see why im so paranoid about parens everywhere! they'll getcha :)

[ladyada]

@earlephilhower thank you for the fix!

[earlephilhower]

Not a problem, it was my mistake to begin with... And I'd say the learning from this is that while parenthesis are important, #define macros with parameters are probably the wrong answer to whatever problem you're having!