[PATCH 1/4] opcodes: use CGEN_INSN_LGUINT for base instructions

[Jose E. Marchesi]

> On 5/17/23 07:03, Jose E. Marchesi via Binutils wrote:
>> 
>> Hi Nick.
>> 
>>> Hi Jose,
>>>
>>>> This patch changes the opcodes CGEN support code in order to allow
>>>> base instructions with opcodes past the least significative 32 bits.
>>>> Note that the masks have been adapted in a previous patch.
>>>> This patch has been regtested for all the current targets in
>>>> binutils
>>>> that are based on CGEN, namely:
>>>> - bpf-unknown-none
>>>> - lm32-elf
>>>> - fr30-elf
>>>> - ip2k-elf
>>>> - iq2000-elf
>>>> - m32c-elf
>>>> - m32r-elf
>>>> - mep-elf
>>>> - mt-elf
>>>> - or1k-elf
>>>> - stormy16-elf"
>>>> Also with --enable-cgen-maint and --enable-targets=all.
>>>> No regressions observed.
>>>
>>> Patch approved - please apply.
>> 
>> Thanks.
>> 
>> Before pushing I am looking at the CGEN_INSN_BYTES point raised by Jan,
>> which made me realize that, even when I am pretty sure the path
>> introduces no regressions, it may be incomplete because it won't cover
>> targets that would have opcodes past the lowest 32-bits of the base
>> instruction and that also define CGEN_INT_INSN_P (even if we currently
>> don't have such targets.)
>
> With the recent changes to the BPF port to not use CGEN, this patch
> is no longer needed, right Jose?

Correct.  To no have to deal with these kind of tangents was a big
motivation for the desCGENization of the port.