GitLab

The realloc() wrapper was included in the develop branch of pico-sdk, so
use it instead of a local, unreproducible version of the SDK.

Should have no effect on code.

* Fix link display in PlatformIO Docs, Remove JLink warning

* Remove unfinished sentence

* Shift arduino attribute higher in JSON file

* Try out PlatformIO CI

* Trigger CI

* Clone recursively, actually use own repo

* Fix YAML

* Build Verbose

* Revert to checkout@v2 version, add TinyUSB to examples

* Try fix TInyUSB example

* Pull repo recursively

* Use v3 after all, correct path to example

* Only do CI on PR

* Update board generation, use renamed function

* Update documentation with new platform integration state

* Remove accidentally pushed file

* Use correct update command

* Use correct highlighting

* Use correct language

* Add section on debugging

* Add docs on filesystem, minor corrections

* Use -iprefix in compilation, but still expose all include paths to IDE

* Add exception and RTTI support, document them

* Fix typo

The makever script sets up the repo properly, so no need to patch things
in the package stage.

Remove MIDI support from the core's main TinyUSB.  MIDI is still supported
using the Adafruit TinyUSB library, just like before (the core never
did use it).

Compile the Pico-SDK using `-Os`

Remove unneeded warning in UF2 upload

Blink.ino shows a savings of 3.5KB flash and 400 bytes of RAM.

For comparison, with this PR blink.ino reports:
````
Sketch uses 49908 bytes (3%) of program storage space. Maximum is 1568768 bytes.
Global variables use 7024 bytes (2%) of dynamic memory, leaving 255120 bytes for local variables. Maximum is 262144 bytes.
````

As comparison, using the MBED core blink.ino reports:
````
Sketch uses 78882 bytes (0%) of program storage space. Maximum is 16777216 bytes.
Global variables use 42780 bytes (15%) of dynamic memory, leaving 227556 bytes for local variables. Maximum is 270336 bytes.
````

So, with this PR we use **37% less flash** and and **87% less RAM**
for simple sketches.

This PR includes a pico-sdk built using the fix in
https://github.com/raspberrypi/pico-sdk/pull/864

To properly fix it requires a new pico-sdk release. but until then I am
building against my own fork of pico-sdk and including the binary.
When 1.3.2 is out (and assuming my PR is approved) I'll update the pico-sdk
link.

Fixes #614

Fixes a hang found while debugging #614.  Do all memory allocations and
USB descriptor setup in main code prior to `tusb_init()`.  Avoids potential
deadlock in cases where the app is allocating while the USB port is being
set up.

Fixes #620

When get.py is run in a script the percent-update printouts shown while
downloading the toolchain end up as 100s to 1000s of lines in log files.

When stdout is not a terminal, avoid printing these percentages and
shrink logfiles significantly. Errors/etc. are still reported as normal.

As noted in #615

Free up 4K of RAM and 6K of flash when no exceptions are enables (default).
The original toolchain was including exception code in libstdc++ by default.

Reduce installation by ~50MB and download by ~25M by removing unused LTO
gcc support.

Fixes #609 and a niggling slowness in uploads.  Try to find the drive
in a loop for 10 seconds, instead of only checking once after 10 seconds.
Avoid 100% CPU usage while waiting for Pico drive

Depends on new toolchain from
https://github.com/earlephilhower/pico-quick-toolchain/pull/11

Fixes #251

Slight difference to generic RP2040
PIN_SERIAL2_TX + PIN_SERIAL2_RX are
switched with PIN_WIRE0_SDA + PIN_WIRE0_SCL

Add plumbing to allow `Wire`, `Serial1`, `SPI1` to map to the 2nd
hardware unit for devices where the PCB layout only brings out the
2nd port.

Fix the Seeedstudio XAIO pins

Fixes #594

- added WIZnet WizFi360-EVB-Pico
- added WIZnet W5500-EVB-Pico
- corrected WIZnet W5100S-EVB-Pico PID

Avoids returning 2^32 bytes in the case of a requestFrom read fails
due to timeout.

Fixed #601

Allows apps to provide a callback parameter in the `attachInterrupt` call
to be passed back to the function when the interrupt happens.

Co-authored-by: Arya11111 <xuepeng@dfrobot.com>

Fixes #593

When SerialPIO::end is called, stop the PIO SMs and potentially
disable the IRQ handler if this is the last RX port on that PIO.

Looks like only minor changes to the SDK, should not affect the core.