This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: [RFC] Add a second 'pinfo' member to mips_opcode to extend number of available bits
- From: Thiemo Seufer <ica2_ts at csv dot ica dot uni-stuttgart dot de>
- To: Fred Fish <fnf at specifixinc dot com>
- Cc: binutils at sources dot redhat dot com
- Date: Mon, 21 Jun 2004 00:03:08 +0200
- Subject: Re: [RFC] Add a second 'pinfo' member to mips_opcode to extend number of available bits
- References: <200406201442.19416.fnf@specifixinc.com>
Fred Fish wrote:
> This patch is really two patches together, which I'll be happy to
> separate into separate sequential patches, if anyone thinks that
> is really necessary.
>
> (1) For the mips binutils there are a couple of INSN_* defines that
> are not used (set to zero) because there are no more bits available in
> the mips_opcode struct member 'pinfo'. Since I wanted to add another
> INSN_* define, I decided to fix this situation at the same time, by
> adding a second struct member for these bits and any new ones needed
> in the future. This patch is pretty large simply because this adds
> an initializer for every entry in the opcode tables.
>
> (2) The mips opcode tables contain entries that look and taste like
> all the other entries, but are really there just to make some special
> cases of a general instruction more logical from a programmer point of
> view. For example, 'nop' is just a special case of an 'sll'
> instruction; 'daddu', 'addu', and 'or' with a zero input operand can
> all be thought of as a 'move', etc. For the purposes of this patch
> I've called these transformations 'aliases', as distinct from 'macros'
> which are handled differently. Some programmers prefer to see the
> actual non-alias instruction in the disassembly output. So I added a
> 'no-aliases' option for objdump's -M to use.
[snip]
> Comments on the option name, patch implementation, etc?
Ok.
Thiemo