		Search-engine friendly clone of the ACL2 documentation.

X86-push-segment-register

PUSH Segment Register

Signature
(x86-push-segment-register proc-mode start-rip temp-rip prefixes rex-byte opcode modr/m sib x86) → x86
Returns: x86 — Type (x86p x86), given (x86p x86).

Note that PUSH CS/SS/DS/ES are invalid in 64-bit mode. Only PUSH FS/GS are valid in 64-bit mode.

0E: [PUSH CS]

16: [PUSH SS]

1E: [PUSH DS]

06: [PUSH ES]

0F A0: [PUSH FS]

0F A8: [PUSH GS]

If the source operand is a segment register (16 bits) and the operand size is 64-bits, a zero-extended value is pushed on the stack; if the operand size is 32-bits, either a zero-extended value is pushed on the stack or the segment selector is written on the stack using a 16-bit move. For the last case, all recent Core and Atom processors perform a 16-bit move, leaving the upper portion of the stack location unmodified.

For now, our model handles the last case described above by doing a 16-bit move. This should be how all modern processor work. In the future, we might parameterize the model on a flag that says how this case is handled (modern or legacy).

PUSH doesn't have a separate instruction semantic function, unlike other opcodes like ADD, SUB, etc. The decoding is coupled with the execution in this case.

Definitions and Theorems

Function: x86-push-segment-register

(defun x86-push-segment-register
       (proc-mode start-rip temp-rip
                  prefixes rex-byte opcode modr/m sib x86)
 (declare (xargs :stobjs (x86)))
 (declare (type (integer 0 4) proc-mode)
          (type (signed-byte 48) start-rip)
          (type (signed-byte 48) temp-rip)
          (type (unsigned-byte 52) prefixes)
          (type (unsigned-byte 8) rex-byte)
          (type (unsigned-byte 8) opcode)
          (type (unsigned-byte 8) modr/m)
          (type (unsigned-byte 8) sib))
 (declare (ignorable proc-mode start-rip temp-rip
                     prefixes rex-byte opcode modr/m sib))
 (declare (xargs :guard (and (prefixes-p prefixes)
                             (modr/m-p modr/m)
                             (sib-p sib)
                             (rip-guard-okp proc-mode temp-rip))))
 (let ((__function__ 'x86-push-segment-register))
  (declare (ignorable __function__))
  (b* ((?ctx 'x86-push-segment-register))
    (b*
      (((the (integer 1 8) operand-size)
        (select-operand-size proc-mode
                             nil rex-byte nil prefixes t t nil x86))
       (rsp (read-*sp proc-mode x86))
       ((mv flg new-rsp)
        (add-to-*sp proc-mode rsp (- operand-size)
                    x86))
       ((when flg)
        (!!fault-fresh :ss 0 :push flg))
       ((the (unsigned-byte 16) val)
        (seg-visiblei (case opcode
                        (14 1)
                        (22 2)
                        (30 3)
                        (6 0)
                        (160 4)
                        (t 5))
                      x86))
       ((mv flg x86)
        (wme-size-opt proc-mode
                      (if (= operand-size 4) 2 operand-size)
                      (the (signed-byte 48) new-rsp)
                      2 val (alignment-checking-enabled-p x86)
                      x86
                      :mem-ptr? nil))
       ((when flg)
        (!!ms-fresh :wme-size-opt flg))
       (x86 (write-*sp proc-mode new-rsp x86))
       (x86 (write-*ip proc-mode temp-rip x86)))
      x86))))

Theorem: x86p-of-x86-push-segment-register

(defthm x86p-of-x86-push-segment-register
  (implies (x86p x86)
           (b* ((x86 (x86-push-segment-register
                          proc-mode start-rip temp-rip prefixes
                          rex-byte opcode modr/m sib x86)))
             (x86p x86)))
  :rule-classes :rewrite)