[PATCH 1/2] x86/Intel: make sure MOVSD/CMPSD have their Size32 honored

[Jan Beulich]

On 07.11.2019 22:52,  H.J. Lu  wrote:
> Another problem is Intel syntax sets instruction suffix from operand size:

An unrelated one, but yes.

> [hjl@gnu-cfl-1 gas]$ cat 2.s
> .intel_syntax noprefix
> call WORD PTR [rax]
> [hjl@gnu-cfl-1 gas]$ as -mintel64 -o 2.o 2.s
> 2.s: Assembler messages:
> 2.s:2: Error: invalid instruction suffix for `call'
> [hjl@gnu-cfl-1 gas]$
> 
> It is operand size mismatch, not invalid instruction suffix.  In case
> of CMPSD and
> and MOVSD,the 'l' suffix is invalid, not DWORD on memory operand.  But suffix
> check treats them the same.  This makes Intel syntax more complicated.   Should
> Intel syntax disallow suffix in mnemonic?  It will make Intel syntax
> closer to SDM.

There are some suffixes that need honoring. The original authors
of the code apparently decided that it's better to accept some
stray/bogus suffixes than to further complicate the code. I guess
once we have the No_*Suf attributes put straight (as mentioned by
Michael the other day, matching plans I've been having), then we
could see about further cleaning up Intel syntax behavior here.

Jan