[PATCH] RISC-V: Mark fgt.*/fge.* as instruction alias

[Nelson Chu]

Thanks for giving the comments :-)  I think we just keep the order, and
then mark them to INSN_ALIAS correctly, even though we don't dump those
aliases, but that just the dis-assemble issue, so we may find a better way
in the future.

Thanks again, committed.
Nelson

On Tue, Apr 29, 2025 at 4:04 PM Jan Beulich <jbeulich@suse.com> wrote:

> On 29.04.2025 10:03, Jan Beulich wrote:
> > On 29.04.2025 09:28, Nelson Chu wrote:
> >> Umm I think it's similar to the
> >> commit 839189bc932ea02c9647a3ad829dda72f8a9562c,
> >> RISC-V: re-arrange opcode table for consistent alias handling
> >>
> >> If we mark the instruction as the alias, then it should be moved to the
> >> top, and then prioritize dump to the alias.  Otherwise should pass
> >> -Mno-aliases to the dis-assembler.  So if fgt is the alias of flt, then
> we
> >> probably should dump flt to fgt by default.
> >
> > That's also an option, but I think the two kinds of aliases (assembler-
> > only vs disassembler-too) warrant the slightly different arrangement in
> > the opcode table for cases like the ones here.
>
> Then again the difference may also warrant leaving things as they are, as
> whether an entry is an alias is relevant to the disassembler only.
>
> Jan
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://sourceware.org/pipermail/binutils/attachments/20250430/e8eb9730/attachment.htm>