Re: [PATCH 1/2] target/m68k: Clear mach in m68k_cpu_disas_set_info

[Richard Henderson]

On 5/17/22 07:06, Laurent Vivier wrote:
> Le 30/04/2022 à 19:02, Richard Henderson a écrit :
> > Zero selects all cpu features in disas/m68k.c,
> > which is really what we want -- not limited to 68040.
>
> But what happens when an instruction has to be decoded differently between 680x0 and 
> coldfire?
>
> for instance in disas/m68k.c, we have:
>
> {"addil", 6,    one(0003200),   one(0177700), "#l$s", m68000up },
> {"addil", 6,    one(0003200),   one(0177700), "#lDs", mcfisa_a },
>
> {"addl", 6, one(0003200),   one(0177700), "#l$s", m68000up },
> {"addl", 6, one(0003200),   one(0177700), "#lDs", mcfisa_a },
>
> {"andil", 6,    one(0001200),   one(0177700), "#l$s", m68000up },
> {"andil", 6,    one(0001200),   one(0177700), "#lDs", mcfisa_a },
>
> {"andl", 6, one(0001200),   one(0177700), "#l$s", m68000up },
> {"andl", 6, one(0001200),   one(0177700), "#lDs", mcfisa_a },
>
> {"bchg", 4, one(0004100),   one(0177700), "#b$s", m68000up },
> {"bchg", 4, one(0004100),   one(0177700), "#bqs", mcfisa_a },
>
> {"bclr", 4, one(0004200),   one(0177700), "#b$s", m68000up },
> {"bclr", 4, one(0004200),   one(0177700), "#bqs", mcfisa_a },
>
> {"bset", 2, one(0000700),   one(0170700), "Dd$s", m68000up | mcfisa_a },
> {"bset", 2, one(0000700),   one(0170700), "Ddvs", mcfisa_a },
> {"bset", 4, one(0004300),   one(0177700), "#b$s", m68000up },
> {"bset", 4, one(0004300),   one(0177700), "#bqs", mcfisa_a },
>
> {"btst", 4, one(0004000),   one(0177700), "#b@s", m68000up },
> {"btst", 4, one(0004000),   one(0177700), "#bqs", mcfisa_a },
>
> {"cmpib", 4,    one(0006000),   one(0177700), "#b@s", m68000up },
> {"cmpib", 4,    one(0006000),   one(0177700), "#bDs", mcfisa_b },
> {"cmpiw", 4,    one(0006100),   one(0177700), "#w@s", m68000up },
> {"cmpiw", 4,    one(0006100),   one(0177700), "#wDs", mcfisa_b },
> {"cmpil", 6,    one(0006200),   one(0177700), "#l@s", m68000up },
> {"cmpil", 6,    one(0006200),   one(0177700), "#lDs", mcfisa_a },

They're all compatible encodings, it's just that the CF ones are more restricted.  When 
debugging a SIGILL on CF, it would be nicer to see

        addl    a0@, d0

than

        .byte   xx, yy

in the -d in_asm dump.

r~