This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH 5/6] x86/MPX: fix operand size handling
- From: "H.J. Lu" <hjl dot tools at gmail dot com>
- To: Jan Beulich <JBeulich at suse dot com>
- Cc: kirill dot yukhin at intel dot com, Binutils <binutils at sourceware dot org>
- Date: Wed, 9 Oct 2013 08:51:23 -0700
- Subject: Re: [PATCH 5/6] x86/MPX: fix operand size handling
- Authentication-results: sourceware.org; auth=none
- References: <5254349502000078000F9A3D at nat28 dot tlf dot novell dot com> <5254364802000078000F9A5D at nat28 dot tlf dot novell dot com> <CAMe9rOpB9=0rgvE0nOZ8o-C1R_EMamgGm8-BRB5DtOGYgs7V0Q at mail dot gmail dot com> <5255239602000078000F9DE5 at nat28 dot tlf dot novell dot com>
On Wed, Oct 9, 2013 at 12:36 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 08.10.13 at 17:45, "H.J. Lu" <hjl.tools@gmail.com> wrote:
>> On Tue, Oct 8, 2013 at 7:43 AM, Jan Beulich <JBeulich@suse.com> wrote:
>>> All MPX instructions in 64-bit mode ignore REX.W, which means we neither
>>> need to encode this bit nor should disassemble with 32-bit register
>>> operands.
>>>
>>> No MPX instructions would ever take a 16-bit register operand.
>>>
>>> gas/
>>> 2013-10-08 Jan Beulich <jbeulich@suse.com>
>>>
>>> * tc-i386.c (process_suffix): Warn about 32-bit register operands
>>> to MPX instructions in 64-bit mode.
>>
>> I think it should be an error.
>
> I can certainly change that - a warning just seemed a better match
> to hardware ignoring operand size here.
We can use separate entries with Reg32 for CpuNo64 and Reg64 for Cpu64,
similar to mov with debug registers. Let's do that instead.
>>> opcodes/
>>> 2013-10-08 Jan Beulich <jbeulich@suse.com>
>>>
>>> * i386-dis.c (intel_operand_size): Move v_bnd_mode alongside the
>>> default case.
>>> (OP_E_register): Move v_bnd_mode alongside m_mode.
>>> * i386-opc.h (REGNAM_RAX): New.
>>> * i386-opc.tbl (bndcl, bndcu, bndcn): Drop Reg16. Add NoRex64.
>>> * i386-tbl.h: Re-generate.
>>>
>>
>> Should we also remove Disp16?
>
> Yes, we certainly could, but then also from the other MPX
> instructions. Logically this would belong into the address size fix
> patch though.
>
> Jan
>
--
H.J.