Re: [PATCH] Lexra support in binutils

[Adam Nemet <anemet at caviumnetworks dot com>]

Maciej W. Rozycki writes:
> On Fri, 24 Oct 2008, Adam Nemet wrote:
> 
> > > I wonder how this will play with .set mips0. Maybe we should use "mips0.9"
> > > for it, "mips0.8" for the compatible octeon subset, and "mips1.9" for the
> > > R5900. :-)
> > 
> > If I understand this correctly, shouldn't octeon be mips64r1.9 then?
> 
>  Oh, we're not short of namespace here; it can be safely declared 
> mips63r2.

My concern was not the actual name but to point out that octeon with the
custom unaligned instructions will have to be defined using ISA63R2 which will
be defined using ISA63 which will be defined using ISA31r2, etc.

Also on the patch specifically:

@@ -170,28 +173,28 @@ const struct mips_opcode mips_builtin_opcodes[] =
 /* name,    args, match,       mask,	       pinfo,		      pinfo2,		membership */
 {"pref",    "k,o(b)",   0xcc000000, 0xfc000000, RD_b,		      0,			   I4|I32|G3	},
 {"prefx",   "h,t(b)",	 0x4c00000f, 0xfc0007ff, RD_b|RD_t,					   0,			I4|I33 },
-{"nop",     "",         0x00000000, 0xffffffff, 0,						   INSN2_ALIAS,		I1      }, /* sll */
+{"nop",     "",         0x00000000, 0xffffffff, 0,						   INSN2_ALIAS,		I1|I0	}, /* sll */

I think these should only belong to I0 and I1 should include I0.

Adam