[PATCH] x86: improve/fix string instruction operand handling
H.J. Lu
hjl.tools@gmail.com
Thu Aug 28 14:24:00 GMT 2008
On Wed, Aug 27, 2008 at 9:23 AM, Jan Beulich <jbeulich@novell.com> wrote:
> All addressing forms were silently accepted for string instructions. While
> most other assemblers appear to also do so, warn about them now. Also, fix
> silent generation of bad code when a displacement is present in one of the
> operands.
>
> Built and tested on i686-pc-linux-gnu.
>
> gas/
> 2008-08-27 Jan Beulich <jbeulich@novell.com>
>
> * config/tc-i386.c (md_assemble): Force number of displacement
> operands to zero when processing string instruction.
> (i386_index_check): Special-case string instruction operands. Don't
> fudge address prefix if there already was a memory operand. Fix
> error message to correctly reflect the addressing mode used.
> (i386_att_operand): Fix comment.
> (i386_intel_operand): Snapshot, clear, and restore base and index
> reg for each operand processed. Increment count of memory operands
> later.
>
> gas/testsuite/
> 2008-08-27 Jan Beulich <jbeulich@novell.com>
>
> * gas/i386/string-bad.{l,s}, gas/i386/string-ok.{d,e,s}: New.
> * gas/i386/i386.exp: Run new tests.
>
It is OK.
Thanks.
--
H.J.
More information about the Binutils
mailing list