exxos's Atari, Amiga & retro forum

Excellent

I managed to port report descriptor parser.
My controllers do work with them, so all controllers which would follow standards should be supported (I wish!).
I most likely will need to cover xbox360 controllers differently.

Next step it to set up report size dynamically, and parse the report and stick it to 8 byte block for sending to cpls.

I am on 81% of flash size used... 64kb ,might be on low side if I need to hold a lot of mapping data for controllers.

I keep by repository here it is for stm32f730, I will port it to f105 when I am done with game-pad support feature.
https://github.com/arkadiuszmakarenko/TFCD32Riser

TBH given the price difference is so small sod it... let’s use the bigger chip and get better support

Next one in QFP64 is STM32F722RCT6 £5.4 + VAT. FS and HS USB ports. 256flash/256ram should be enough.
It looks like for our use it would be pin compatible with f105/f730 with decent margin for potential growth of this project.
I will order 2 for tests in next few days.

Honestly.. if you get more devices working I think it justifies the cost

Ok. Report parser ported and output packed in 1byte.

This conclude initial work on gamepad support.
Next composite devices.

Awesome progress. Well done. :)

A little update on progress.
I managed to butcher and rebuild library make it work with multiple interface configurations.
After battling with overflow of one of buffers to my precious usb state machine and corrupting first two states, I managed to receive both data from keyboard and mouse at the same time.
Issue is that in this scenario mouse refuses to work in boot mode.(data cannot have the same reportId after all). Library is hardcoded only to request descriptor for interface index 0, so cannot read automatically report at this moment and make sense of data.
I need to learn how to build requests for descriptors, update few things, fix report descriptor parser to support every mouse.
Fortunately keyboard works exactly the same.
So yes, I should be able to provide support for some composite devices, but it will take some time.
There is still a lot of work to do, but I am happy that I managed to prove the concept of library structure.

"Issue is that in this scenario mouse refuses to work in boot mode.(data cannot have the same reportId after all). Library is hardcoded only to request descriptor for interface index 0, so cannot read automatically report at this moment and make sense of data."

Got the fcker! A LOT of hardcoded values, very easy to miss this kind of stuff!

@terriblefire

Do we really want hub support for this project?