[J-core] Illegal instruction handling (was: Re: PC-relative loads and delay slots)

[Rob Landley]

On 07/20/2016 05:13 AM, Robert Ou wrote:
> On Tue, Jul 19, 2016 at 11:40 PM, Robert Ou <rqou at robertou.com> wrote:
> Incidentally, I may have discovered a bug in binutils. sh2elf-objdump
> (Rob Landley's binary version) disassembles 0x0048 as clrs and 0x0058
> as sets even when given the -msh2 option. These instructions were
> (according to http://www.shared-ptr.com/sh_insns.html) only introduced
> in SH3.

My toolchain is obsolete, between the the 4.4 kernel and current vanilla
binutils 2.17 stopped building 4 different architectures (arm, mips,
sh4, and powerpc), and I don't have time to keep up with the whack-a-mole.

Rich recently extended his toolchain build
(https://github.com/richfelker/musl-cross-make) to bulid native
compilers, and I have a todo item to make my aboriginal build work with
the output of his build as its toolchains.

But right now AOSP is in the "Android O" merge window (AOSP is the
android "upstream" that they fork to create each new android release),
and I'm trying to get as many toybox commands ready so they can switch
over to the toybox version for Android O. (Elliott Hughes, the Android C
library and command line maintainer, posted a list of remaining
non-toybox commands a couple weeks back at
http://lists.landley.net/pipermail/toybox-landley.net/2016-June/008484.html
and I'm working through said list.)

tl;dr: use Rich's toolchain instead of mine. I plan to start doing that
myself in future.

Rob

(I'm not posting gplv3 binaries because I consider it a proprietary
license and don't want to get sued, but Rich might post them somewhere?
He hasn't been involved in as many IP lawsuits as I have...)