micropython: update to 1.14

[lfdebrux]

General

• This is a new package and it conforms to the quality requirements

Have the results of the proposed changes been tested?

• I use the packages affected by the proposed changes on a regular basis and confirm this PR works for me
• I generally don't use the affected packages but briefly tested this PR

Does it build and run successfully?

(Please choose at least one native build and, if supported, at least one cross build. More are better.)

• I built this PR locally for my native architecture, (x86_64-musl)
• I built this PR locally for these architectures (if supported. mark crossbuilds):
  • aarch64-musl
  • armv7l
  • armv6l-musl
  • armv5tel-musl

[ericonr]

Please squash the commits into a single one as well :)

[ericonr]

This might (almost definitely) be a bug you want to report to musl folks. Can you show a stack trace / disassembly output?

[ericonr]

@lfdebrux can you check if their build system doesn't simply disable PIC? By passing -fno-PIC to the compiler, for example. If that's the case, what can happen is that it interacts badly with our PIE by default toolchain, and creates a text relocation in the binary. Musl doesn't support those at all.

You can check for their presence with readelf -d <binary> and seeing if TEXTREL appears in the output.

[lfdebrux]

It doesn't look like they disable PIC; what's weird is that it works fine on x86_64-musl with PIE, but doesn't on armv5tel-musl, which as you say does point towards it being a bug in musl.

Here's a stacktrace from gdb

(gdb) start
Temporary breakpoint 1 at 0x90f0: file main.c, line 421.
Starting program: /usr/bin/micropython

Program received signal SIGSEGV, Segmentation fault.
0xb6fb68c0 in do_relocs () from /lib/ld-musl-arm.so.1
(gdb) backtrace
#0  0xb6fb68c0 in do_relocs () from /lib/ld-musl-arm.so.1
#1  0xb6fb6c88 in reloc_all () from /lib/ld-musl-arm.so.1
#2  0xb6fb900c in __dls3 () from /lib/ld-musl-arm.so.1
#3  0xb6fb867c in __dls2 () from /lib/ld-musl-arm.so.1
#4  0xb6fb569c in _dlstart () from /lib/ld-musl-arm.so.1
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb)

[ericonr]

Seems it also segfaults on armv6l :P

I will try to determine which platforms we have to enable nopie for.

[ericonr]

0x00000016 (TEXTREL) 0x0

As suspected :) So not really a bug on musl's side, though it would be nice if an error message were added.

This might be due to their assembly code somewhere, though I couldn't find it. Weirdly, Alpine disables armv7l as a target because tests don't pass, and even armhf (which is equivalent to our armv6l) because it doesn't build (it might simply be outdated, of course).

[lfdebrux]

Oh interesting. With nopie that isn't present. Sorry I can't be more helpful, I'm a bit out my depth!