Alessandro Gatti

Apologies for the extra push, it's just a fix for a minor issue I found when reviewing the code for the last time (the check for negative values' range didn't...

> * use `qstr_short_t` for the type of `REGISTERS_QSTR_TABLE` Oh, I don't think I've ever seen `qstr_short_t` being used before. So I must assume there's a hard limit of 64k...

For `features2` and `random` I have no idea at the moment on how to tweak the build process to get the soft-float code work on Picolibc in the CI environment....

@dpgeorge I found an "interesting" bug, which I'm not sure if it is caused by my relocation handling. This happens when you build the `examples/deflate` natmod for rv32imc and then...

The `framebuf` natmod crashes because `pyelftools` may have a bug in handling `R_RISCV_SUB32` relocations. The `switch` statement in `framebuf_make_new` is compiled as a table whose entries contains two relocations each,...

@dpgeorge here's a version that makes all tests pass for the natmods that build (`features0`, `features1`, `features3`, `features4`, `framebuf`, `heapq`, `re`) except for the ones in `deflate` and `framebuf_ellipse` and...

Can reproduce the same crashes on `xtensawin` and an ESP32-D0WD-V3.

> Everything passes locally for me, except compiling `features2` (because it needs float) and running `extmod/random_extra_float.py` (because it needs float). If you merge #15838 first then RV32 natmods requiring floating...

@dpgeorge I guess you've tested building modules using newlib outside the CI? The `btree` natmod was never able to be built with picolibc due to errno-related incompatibilities. I believe the...

> I'm using the `riscv32-embecosm-ubuntu2204-gcc13.2.0` toolchain (used for the Pico 2) and I guess that uses newlib? So far the only picolibc-only RV32/RV64 toolchain I've found is the one shipping...