Re: Bug in operands to IA32 cvtps2dq instruction in gnu as

["H . J . Lu" <hjl at lucon dot org>]

On Sat, May 12, 2001 at 09:16:40PM +0930, amodra@one.net.au wrote:
> This should fix twobyte_has_modrm, and ring bells in the future when
> people add new opcodes without updating the table.  Makes your patch
> redundant H.J., and in any case it wasn't a fix except when the modrm
> byte just happenned to be zero.

I am not sure about making it file scope. Is that really what we
want? I thought it was set by the current insn and you had to check
it for each insn. Otherwise, you will check the modrm byte even if
it is not there and just because the previous insn has it.

H.J.