embassy+ufmt+usart: unexpected outputs?

[brainstorm]

I'm confused with the Usart+ufmt behavior on my Atmega2560, see this output in the commit message: brainstorm/embassy-as5600@a37b9b5

Probably an embassy main task related issue since it should print just one line but seems to print garbage in a loop instead?

The baudrate seems fixed to 115200 everywhere (flashing (in .cargo/config.toml with ravedude) and running via cargo run)... it also looks like a baudrate mismatch issue but I cannot see where the different baudrate setting can come from? Here's my .cargo/config.toml:

[build]
target = "avr-specs/avr-atmega328p.json"

[target.'cfg(target_arch = "avr")']
runner = "ravedude mega2560 -cb 115200"

[unstable]
build-std = ["core"]

My experience looks very similar to #466, but my baudrates seem consistent and set to 115200 in the code itself too?

/cc @djdisodo

[brainstorm]

I just added support for Wokwi in commit brainstorm/embassy-as5600@4db8d66 (kudos to @Vollbrecht for #415) and while it prints something a bit closer to Meow World! it's not that exact string, but this instead Me w Wr d (see screenshot below) ... those series of characters exactly twice per simulation run:

Very interesting! 🤔 Here's the snippet/project hosted at Wokwi if you want to examine it there: https://wokwi.com/projects/393953834443965441

[brainstorm]

At 9600 I get the same output, this time for the Wokwi terminal instead of the VSCode serial port extension, as suggested by Wokwi docs... just to make sure it's not an artifact of neither the VSCode serial port extension nor the baudrate value itself:

I guess next steps are examining which assembly is generated or do you have other ways to debug this more effectively, @Rahix?

[stappersg]

Acknowlegde on "baudrates match", but how about "chip match"?

I would expect both AVR devices being the same.

[brainstorm]

Ouch, that was embarrassing, thank you so much, works now!

[Rahix]

Glad you got it figured out - these little mistakes can be very hard to spot...

Although my takeaway is that ravedude really should be safeguarding against this. I opened #531 for the topic.