Fix some regressions on feather M7

[jepler]

This closes #7944 but it's fishy -- why is such a large buffer required? @hathach your feedback would be appreciated about this question.

It also includes a board-specific fix for #7952 but this may be needed on other boards. @tannewt do you know what's going on here? The USB_DM/USB_DP pins seem to just be dummy objects on 1011 but adding these to the list (as you did on the 1011 evk) does allow CP to start. If this is correct, the fix should be propagated to ALL boards, or incorporated at a better location.

ports/mimxrt10xx/peripherals/mimxrt10xx/MIMXRT1011/pins.c:
const mcu_pin_obj_t pin_USB_OTG1_DN = { { &mcu_pin_type }, };

[tannewt]

@tannewt do you know what's going on here? The USB_DM/USB_DP pins seem to just be dummy objects on 1011 but adding these to the list (as you did on the 1011 evk) does allow CP to start. If this is correct, the fix should be propagated to ALL boards, or incorporated at a better location.

I don't. I didn't think that the reset code would get this far because it should be bounded by the IOMUXC count. Must be some loop bounds error.

[hathach]

This closes #7944 but it's fishy -- why is such a large buffer required? @hathach your feedback would be appreciated about this question.

because usb highspeed endpoint size for bulk is 512 bytes. Maybe the buffer is overflowed or something. I will try to pull out and test tomorrow to see if we could still shrink the memory and not get into trouble.

[jepler]

@tannewt can we merge this even if the reason the USB_DM/DP change fixes anything is murky? It's unfortunate and frustrating that metro m7 seems to simply not work at present.

[gamblor21]

Tested and this did give me access back to the Metro M7. If things work out I will get a new MIDI keyboard tomorrow and could try the MIDI part.

[jepler]

#7964 was the better fix for the not-booting-at-all problem. This is now just to fix the MIDI problem. I only increased the RX (to microcontroller) buffer, and only tested this direction. Before merging, both directions should be tested to find out if they work.

[jepler]

In my testing, the M7 can now receive (from linux vkeybd) and send (to linux timidity) midi events. So, the send buffer doesn't currently need a size increase, just the receive buffer.

[tannewt]

Thank you!