@h(IBM XA Code Planting Routines)

The following routines have been provided initially, corresponding to the fo
formats given in the "@b(Principles of Operation)".

   @B(PI RR) (%integer @b(Op, R1, R2))

   @b(PI RRE)(%integer @b(Op, R1, R2))

   @b(PI RX) (%integer @b(Op, R1, X2, B2, D2))

   @b(PI RS) (%integer @b(Op, R1, R3, B2, D2))

   @b(PI SI) (%integer @b(Op, I2, B1, D1))

   @b(PI S)  (%integer @b(Op, B2, D2))

   @b(PI SS) (%integer @b(Op, L1, L2, B1, D1, B2, D2))

   @b(PI SSE)(%integer @b(Op, B1, D1, B2, D2))


I would welcome suggestions for further routines, changes in name to suit
existing practice, dummy operands to be provided etc.

@n()
@h(jump routines)

After initial discussion of the problems of jumps on IBM architecture I am
proposing the following enhancements to the jump routines in the put interface.

   @a(;) Extend the parameters to PJump to include Mask and Reg. Mask is the
         condition mask or register field. Reg is a work register which is free
         for PJump to use. Both are integers. If the top bit of Reg is set, this
         is the same work register as has just been used in a preceeding PJump.

   @a(;) Add two new routines PUsing and PDrop, to tell the put interface that a
         register has just been used to hold the CA and that it has been reused.
         Both take a single parameter, %integer Reg.

@n()
@h(Prepared files)

Two files have been prepared for use by code generators.

      @b(ERCS12.IBMCODE_MNEMONICS) - a list of constinteger declarations of
            opcodes, based on the XA Principles of Operation, ordered by value.

      @b(ERCS12.IBMCODE_PROPS) - a constinteger array from 0-255, holding
            property information for the single byte opcodes. Currently the
            least significant byte holds the instruction format.

Rob Pooley, 1st March 1984