[PATCH] i386: Check invalid (%dx) usage

[Jan Beulich]

On 10.11.2022 18:22, H.J. Lu wrote:
> On Wed, Nov 9, 2022 at 11:21 PM Jan Beulich <jbeulich@suse.com> wrote:
>>
>> On 09.11.2022 21:24, H.J. Lu wrote:
>>> On Tue, Nov 8, 2022 at 11:21 PM Jan Beulich <jbeulich@suse.com> wrote:
>>>>
>>>> On 08.11.2022 22:06, H.J. Lu wrote:
>>>>> On Mon, Nov 7, 2022 at 11:34 PM Jan Beulich <jbeulich@suse.com> wrote:
>>>>>> On 07.11.2022 20:58, H.J. Lu wrote:
>>>>>>> On Mon, Nov 7, 2022 at 3:44 AM Jan Beulich <jbeulich@suse.com> wrote:
>>>>>>>> x86: restrict use of (%dx)
>>>>>>>>
>>>>>>>> PR gas/29751
>>>>>>>> The AT&T mode special case operand (%dx) is valid to use only with
>>>>>>>> instructions nominally expecting %dx to specify an I/O port address.
>>>>>>>> Prefix the respective checking with an opcode check. Keep that as
>>>>>>>> simple as possible by recognizing that opcodes 0x64 and 0x66 (which
>>>>>>>
>>>>>>> Since current_templates doesn't point to the matched instruction,
>>>>>>> checking current_templates looks like abuse.  I don't think error
>>>>>>> messages should be a concern here.
>>>>>>
>>>>>> We use current_templates in similar ways in quite a number of places,
>>>>>> when match_templates() hasn't run yet.
>>>>>
>>>>> Since the first template isn't the selected one, your check allows
>>>>> the invalid opcodes.
>>>>
>>>> I guess I don't understand, but I guess I'll also give up. Which
>>>
>>> Your proposed change does
>>>
>>> current_templates->start->base_opcode | 0x8a) == 0xee
>>>
>>> to allow opcode 0xe4 and (%dx) is allowed for non-I/O opcodes.
>>
>> 0xe4 is very much an I/O opcode, merely one not allowing for (%dx).
> 
> But it also matches other opcodes.

Let's enumerate them all:

0x64	gs: prefix
0x66	data prefix
0x6c	ins
0x6e	outs
0xe4	in
0xe6	out
0xec	in
0xee	out

As said in the description: "opcodes 0x64 and 0x66 (which wrongly also
match the check) encode prefixes, which hence - even if used standalone -
don't take any operands, so match_template() will fail there for other
reasons."

Jan