This is the mail archive of the mailing list for the binutils project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH 02/10] x86: drop stray W

On 07.08.2019 17:13,  H.J. Lu  wrote:
On Wed, Aug 7, 2019 at 12:43 AM Jan Beulich <> wrote:

On 06.08.2019 21:37,  H.J. Lu  wrote:
On Tue, Aug 6, 2019 at 7:25 AM Jan Beulich <> wrote:

The flag is used to indicate opcodes which can be switched between byte
and word/dword/qword forms (in a "canonical" way). Obviously it's quite
odd then to see it on insns not allowing for byte operands in the first
place. As a result the opcode bytes need to be adjusted accordingly,
which includes comparisons done in optimize_encoding().

These encodings do allow byte operand.

By "encodings" I assume you mean the opcodes, not the templates. The
templates modified here all clearly don't allow byte operands, and
that's what counts when considering whether W is applicable. is checked only in process_suffix.  This part

       /* It's not a byte, select word/dword operation.  */
       if (
           if (
    |= 8;
    |= 1;

applies to encoding.  Even if we can't merge entries in i386-opc.tbl,
W still makes senses.   Will keeping W cause any issues?

Probably not right now, but I'd have to invest time to re-check the
rest of the series without it. But I still don't get it: Other than
what you say, W does _not_ make sense when no accepted operand
combination allows the if() above to be bypassed. It is an "alter
the encoding if the operand is word/dword/qword, i.e. not byte" flag,
implying that the encoding should remain unchanged for byte operands,
which the templates in question don't accept in the first place.

Let me state my position in another way: Every, absolutely every
attribute in the templates should have a reason to be there.
Everything else should be dropped. Over the last couple of years
I've managed to get rid of quite a few pointlessly present
attributes. The W flags here are just another example. Not
following this fundamental way of handling things has led to the
mess that the opcode table was and to a fair degree still is.
This is actively hindering maintainability.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]