The audio output objects all have the same general necessary configuration
calls. Abstract them out to a generic AudioOutputBase interface class that
they will inherit from. Simplifies letting applications use different
output channels.
Should be backwards compatible with existing code.
Semihosting is handy for debugging, so allow the core to use `SerialSemi` as the
::printf port. Add menu item to the IDE to allow selection.
Add RISC-V implementation of semihost call
Adds a menu item to enable onboard profiling. This requires significant
RAM and really only makes sense on devices with PSRAM to store the state.
When the menu item is selected, allocates RAM and tracks function calls and
periodically samples the PC to generate a histogram of application usage.
The onboard gmon.out file can be written over Semihosting or
some other way to transfer to a PC for analysis.
Adds a profiling example with command lines.
Using pico-sdk develop branch, add in support for CYW43-based
WiFi/BT/BLE boards on the RP2350 such as the
SparkFun Thing Plus RP2350 or the Pimoroni Pico Plus 2W.
Fixes#2608
Rolls in dynamic SPI divider #2600
* Support LED digitalWrite on RP2350+CYW
Also move "special GPIO" to 64 since the Pimoroni Pico 2W uses the
RP2350B with 48 GPIOs.
* Enable CYW43_PIN_WL_DYNAMIC in IDE and P.IO
Allows calling `cyw43_set_pins_wl(cyw43_pin_array);` to redefine the
CYW43 hookup in the variant initialization.
The RP2350 boot ROM seems to randomize memory on a HW reset, including
both CPU stacks where we normally stuff the "reset to bootloader" flag.
Update the docs and source to remove rp2040.enableDoubleResetBootloader()
on RP2350-based boards.
Fixes#2606
Remove old pre-release BTStack files from sdkoverrides (were needed for
compound BLE devices, but the latest BTStack now incorporates these
changes).
Fixes#2547
* Workaround PSRAM cache invalid'n by reading flash
Fixes#2537
While waiting for a working direct cache flush routine, try and force the
cache to evict all PSRAM values by reading a bunch of flash addresses (which
share the XIP cache). This hurts performance when PSRAM is not used, but
is required for correctness until we have a working XIP flush.
* Invalidate after cleaning the cache line
gmx from the RPI forums came up with this hack and it seems to work!
https://forums.raspberrypi.com/viewtopic.php?p=2262371#p2262371
33% faster for 4K memcpy using DMAMemcyp example
With this assembly:
CPU: 4835 clock cycles for 4K
DMA: 2169 clock cycles for 4K
Using stock Newlib memcpy:
CPU: 7314 clock cycles for 4K
DMA: 2175 clock cycles for 4K
The flash ROM routines seem to overwrite the QMI configuration we set for
PSRAM, rendering it unreadable after any erase or write or ID command.
Wrap the 4 flash control functions to preserve the QMI state on the
RP2350. On the RP2040, simply pass through the call.
Fixes#2537
* Adds RISC-V compilation option to the IDE and Platform.IO.
* Build RP2350-RISCV libpico, libbearssl
* Fix RP2350 BearSSL library (was copied from RP2040, now built for M33)
* New GCC 14.2 toolchain is required (12.4 RISC-V support is borked)
* Newlib locking fixed prototypes
* Manually force all runtime init code into RP2350 binaries
* Add RISC-V to CI
* Remove RP2350 BOOT2.S files, binaries (not used)
* Clean up minor GCC 14.x warnings
* Add RP2350-RISCV OTA build, link
* Add RISC-V FreeRTOS files (configuration still not running, but builds)
* Add basic documentation
Several Arduino APIs realloc(NULL) which is legal and equivalent to
"malloc()", but the PSRAM logic was placing those malloc calls in PSRAM
and not RAM because "0" < RAM_START.
Ensure the realloc address is non-null and before RAM_START before
using PSRAM.
Pull in Raspberry Pi's custom RP2350 ARM and RISC-V ports for FreeRTOS.
Basic tests run, but stress mutex test is failing in unique and interesting
ways.
* Add simplified switching test catching task swap problem
* Freertosrp2350: use FreeRTOS macros in noInterrupts/interrupts when applicable. (#2456)
* Use FreeRTOS macros in noInterrupts/interrupts when applicable.
* Fixed calling taskEXIT_CRITICAL and taskENTER_CRITICAL
---------
Co-authored-by: fietser28 <fietser28@users.noreply.github.com>
* Add support for the extra 16 GPIO pins in the menus and core.
* Clean up Generic RP2350 PSRAM ("none" is valid) and flash (other than 16MB) options.
* Add extra GPIO<->peripheral connections
* Add Pimoroni PGA2350 RP2350B-based board
* Pins 32-47 can be used for PIOPrograms
* Avoid hang when PSRAM fails to initialize
* Move libpico to an RP2350B board for SDK (otherwise the SDK drops all GPIOHI support)
SWSerial wasn't even building due to a typo in the header, and SerialPIO
needs to set the OE-invert flag after PIO initialization for transmit.
Fixes#2419
Use real GPIO pad inversion to allow inverted RX, TX, and controls for
the hardware UART and software PIO-emulated serial ports.
Adds ``setInvertTX(bool)`` and ``setInvertRX(bool)`` calls to both ports,
with ``setInvertControl(bool)`` for the HW UARTS.
* Migrate RP2040-specific bits to separate dirs
* Add chip to boards.txt, isolate RP2040-specifics
* Add RP2350 boot2, bearssl, and libraries
* Platform.IO adjust to new paths
* Add RPIPICO2 JSON for P.IO
* Add RP2350 to Platform.io
* Update Picotool and OpenOCD for all hosts
* Use picotool to generate UF2s
* Build separate libpico blobs serially
Thanks for the review, @aarturo182 !
* Add RP2350 to CI
* Allow Ethernet/WiFi building for RP2350
* Update Adafruit TinyUSB to latest
* Test skip fix
* Make RP2350 Picotool work. update USB ID
* Fix EEPROM/FS flash locations
RP2350 adds a 4K header sector to the UF2, meaning we have 4K less total
flash to work with. Adjust all constants appropriately on the RP2350.
* Adds ilabs board and PSRAM support. (#2342)
* Adds iLabs boards and basic PSRAM support.
* Make PSRAM come up as part of chip init
Uses SparkFun psram.cpp to set timings on clocks which are defined in the
variant file. Prefix things with RP2350_PSRAM_xxx for sanity.
Users don't need to call anything, PSRAM "just appears". Still need to
add in malloc-type allocation.
* Add board SparkFun ProMicro RP2350
Same pinout as the SparkFun ProMicro RP2040 with 8MB PSRAM and RP2350
* Add TLSF library for use w/PSRAM
Fork of upstream to include add'l C++ warning fixes.
* Add pmalloc/pcalloc to use PSRAM memory
free() and realloc() all look at the pointer passed in and jump to the
appropriate handler. Also takes care of stopping IRQs and taking the
malloc mutex to support multicore and FreeRTOS (when that workd)
* Fix BOOTSEL for RP2350
* Add simple rp2040.idleOtherCore test
* Add Generic RP2350 and clean up PSRAM menus
Commercial boards now only have 1 size PSRAM, no need to have menu for them.
* Add Solder Party RP2350 Stamp boards (#2352)
* Add PSRAM heap info helpers, mutex lock mallinfo
* Add RP2350 docs
* FreeRTOS and OTA unsupported warnings for RP2350
* Update to 2.0.0 SDK
* Board type needs to be set before earliest SDK setup
* Platform includes update
* Boot2 files
* Simple compilation issues
* Build and link
* PIO rebuild with version
* Newlib wrapper update
* Force inclusion of runtime_init_* fcns
The linker was dropping all references to the library's included
runtime_init_xxx functions and hence things like the IRQ vector
table and mutexes and alarms weren't properly set up leading to
instant crashes on start up..
Explicitly call out one function from the object file stored in
the .A to force the inclusion of all the functions. May be a better
way, heppy to hear any ideas.
* Fix SPI GPIO calls
* Fix Ethernet GPIO
* Remove SDK warnings
Remove the skipped error messages once the following PR merged:
https://github.com/raspberrypi/pico-sdk/pull/1786
* BTStack moved SBC encode/decode paths
* Platform.IO fixes
* BT No longer has special absolute mouse
* Rebuild and update OTA
* Rebuild BearSSL, too
* Update liker file to latest SDK
* Clean up libpicocmake
* Clean up LWIP/BT library names
* Add VFS to enable POSIX file I/O operations
Enables use of FILE * operations on internal and external storage. fopen,
fclose, fseek, fprintf, fscanf, etc. supported.
* Add FS/File::stat and support POSIX stat/fstat
Removed FS::info64, and updates FS::info with the 64-bit version since in
2024 it's almost impossible to get a SD card smaller than 4GB.
Most code can simply replace info64 with info and continue operation, if they
were updated to be 64-bit in the first place.
RP2040::memcpyDMA implements a DMA-controlled memory copy call identical in
function to standard memcpy, but using an onboard DMA engine. For large
memory transfers this can be significantly faster than using the CPU-based
memcpy. Only 4-byte aligned source, destination, and counts are allowed.
If any inputs are not 4-byte aligned, then standard memcpy will occur so
it will behave correctly for any inputs.
