Re: [PATCH] fix disassembly of spurious REX prefix before FWAIT

["H.J. Lu" <hjl dot tools at gmail dot com>]

On Wed, Oct 24, 2012 at 10:00 AM, Roland McGrath <mcgrathr@google.com> wrote:
> On Wed, Oct 24, 2012 at 3:32 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> Have you verified that REX is ignored in hardware?
>
> I've verified that the REX prefix is part of the FWAIT instruction rather
> than the FNSAVE instruction (FSAVE is shorthand for FWAIT;FNSAVE).  I can't
> tell what it would mean for the REX prefix to be used or not by FWAIT.  I
> don't know how to test that FWAIT actually does its work when it has a REX
> prefix.  It's certainly the case that single-step at REX FWAIT leaves the
> PC after the FWAIT (at the FNSAVE) just like single-step at plain FWAIT does.
>
>> Please put the new tests in rex.s.
> [...]
>> Please use "fsave  (%r8)" instead.
>
> Done.
>
>
> Thanks,
> Roland
>
>
> gas/testsuite/
> 2012-10-24  Roland McGrath  <mcgrathr@google.com>
>
>         * gas/i386/rex.s: Add test of REX prefix before fsave (i.e. fwait).
>         * gas/i386/rex.d: Update.
>
> opcodes/
> 2012-10-24  Roland McGrath  <mcgrathr@google.com>
>
>         * i386-dis.c (ckprefix): When bailing out for fwait with prefixes,
>         set rex_used to rex.
>

OK,  Thanks.

-- 
H.J.