freepascal.compiler/compiler/x86/nx86add.pas

{
    $Id$
    Copyright (c) 2000-2002 by Florian Klaempfl

    Common code generation for add nodes on the i386 and x86

    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation; either version 2 of the License, or
    (at your option) any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program; if not, write to the Free Software
    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.

 ****************************************************************************
}
{
Common code generation for add nodes on the i386 and x86
}
unit nx86add;

{$i fpcdefs.inc}

  interface

    uses
       node,nadd,ncgadd,cpubase;

    type
       tx86addnode = class(tcgaddnode)
         procedure second_addfloat;override;
         procedure second_addfloatsse;
         procedure pass_left_and_right(var pushedfpu:boolean);
       end;


  implementation

    uses
      globals,
      verbose,
      aasmtai,
      cpuinfo,
      cgbase,cgobj,cgx86,cga,
      pass_2,ncgutil,
      defutil;

{*****************************************************************************
                                AddFloat
*****************************************************************************}

    procedure tx86addnode.pass_left_and_right(var pushedfpu:boolean);
      begin
        { calculate the operator which is more difficult }
        firstcomplex(self);

        { in case of constant put it to the left }
        if (left.nodetype=ordconstn) then
         swapleftright;
        secondpass(left);

        { are too few registers free? }
        if location.loc=LOC_FPUREGISTER then
          pushedfpu:=maybe_pushfpu(exprasmlist,right.registersfpu,left.location)
        else
          pushedfpu:=false;
        secondpass(right);
      end;


    procedure tx86addnode.second_addfloat;
      var
        op         : TAsmOp;
        resflags   : tresflags;
        pushedfpu,
        cmpop      : boolean;
      begin
        if use_sse(resulttype.def) then
          begin
            second_addfloatsse;
            exit;
          end;
        pass_left_and_right(pushedfpu);

        cmpop:=false;
        case nodetype of
          addn :
            op:=A_FADDP;
          muln :
            op:=A_FMULP;
          subn :
            op:=A_FSUBP;
          slashn :
            op:=A_FDIVP;
          ltn,lten,gtn,gten,
          equaln,unequaln :
            begin
              op:=A_FCOMPP;
              cmpop:=true;
            end;
          else
            internalerror(2003042214);
        end;

        if (right.location.loc<>LOC_FPUREGISTER) then
         begin
           cg.a_loadfpu_loc_reg(exprasmlist,right.location,NR_ST);
           if (right.location.loc <> LOC_CFPUREGISTER) and
              pushedfpu then
             location_freetemp(exprasmlist,left.location);
           if (left.location.loc<>LOC_FPUREGISTER) then
            begin
              cg.a_loadfpu_loc_reg(exprasmlist,left.location,NR_ST);
              if (left.location.loc <> LOC_CFPUREGISTER) and
                 pushedfpu then
                location_freetemp(exprasmlist,left.location);
            end
           else
            begin
              { left was on the stack => swap }
              toggleflag(nf_swaped);
            end;

           { releases the right reference }
           location_release(exprasmlist,right.location);
         end
        { the nominator in st0 }
        else if (left.location.loc<>LOC_FPUREGISTER) then
         begin
           cg.a_loadfpu_loc_reg(exprasmlist,left.location,NR_ST);
           if (left.location.loc <> LOC_CFPUREGISTER) and
              pushedfpu then
             location_freetemp(exprasmlist,left.location);
         end
        else
         begin
           { fpu operands are always in the wrong order on the stack }
           toggleflag(nf_swaped);
         end;

        { releases the left reference }
        if (left.location.loc in [LOC_CREFERENCE,LOC_REFERENCE]) then
          location_release(exprasmlist,left.location);

        { if we swaped the tree nodes, then use the reverse operator }
        if nf_swaped in flags then
          begin
             if (nodetype=slashn) then
               op:=A_FDIVRP
             else if (nodetype=subn) then
               op:=A_FSUBRP;
          end;
        { to avoid the pentium bug
        if (op=FDIVP) and (opt_processors=pentium) then
          cg.a_call_name(exprasmlist,'EMUL_FDIVP')
        else
        }
        { the Intel assemblers want operands }
        if op<>A_FCOMPP then
          begin
             emit_reg_reg(op,S_NO,NR_ST,NR_ST1);
             tcgx86(cg).dec_fpu_stack;
          end
        else
          begin
             emit_none(op,S_NO);
             tcgx86(cg).dec_fpu_stack;
             tcgx86(cg).dec_fpu_stack;
          end;

        { on comparison load flags }
        if cmpop then
         begin
           cg.getexplicitregister(exprasmlist,NR_AX);
           emit_reg(A_FNSTSW,S_NO,NR_AX);
           emit_none(A_SAHF,S_NO);
           cg.ungetregister(exprasmlist,NR_AX);
           if nf_swaped in flags then
            begin
              case nodetype of
                  equaln : resflags:=F_E;
                unequaln : resflags:=F_NE;
                     ltn : resflags:=F_A;
                    lten : resflags:=F_AE;
                     gtn : resflags:=F_B;
                    gten : resflags:=F_BE;
              end;
            end
           else
            begin
              case nodetype of
                  equaln : resflags:=F_E;
                unequaln : resflags:=F_NE;
                     ltn : resflags:=F_B;
                    lten : resflags:=F_BE;
                     gtn : resflags:=F_A;
                    gten : resflags:=F_AE;
              end;
            end;
           location_reset(location,LOC_FLAGS,OS_NO);
           location.resflags:=resflags;
         end
        else
         begin
           location_reset(location,LOC_FPUREGISTER,def_cgsize(resulttype.def));
           location.register:=NR_ST;
         end;
      end;


    procedure tx86addnode.second_addfloatsse;
      var
        op : topcg;
      begin
        pass_left_right;
        if (nf_swaped in flags) then
          swapleftright;

        case nodetype of
          addn :
            op:=OP_ADD;
          muln :
            op:=OP_MUL;
          subn :
            op:=OP_SUB;
          slashn :
            op:=OP_DIV;
          else
            internalerror(200312231);
        end;

        location_reset(location,LOC_MMREGISTER,def_cgsize(resulttype.def));
        { we can use only right as left operand if the operation is commutative }
        if (right.location.loc=LOC_MMREGISTER) and (op in [OP_ADD,OP_MUL]) then
          begin
            location.register:=right.location.register;
            { force floating point reg. location to be written to memory,
              we don't force it to mm register because writing to memory
              allows probably shorter code because there is no direct fpu->mm register
              copy instruction
            }
            if left.location.loc in [LOC_FPUREGISTER,LOC_CFPUREGISTER] then
              location_force_mem(exprasmlist,left.location);
            cg.a_opmm_loc_reg(exprasmlist,op,location.size,left.location,location.register,mms_movescalar);
            location_release(exprasmlist,left.location);
          end
        else
          begin
            location_force_mmregscalar(exprasmlist,left.location,false);
            location.register:=left.location.register;
            { force floating point reg. location to be written to memory,
              we don't force it to mm register because writing to memory
              allows probably shorter code because there is no direct fpu->mm register
              copy instruction
            }
            if right.location.loc in [LOC_FPUREGISTER,LOC_CFPUREGISTER] then
              location_force_mem(exprasmlist,right.location);
            cg.a_opmm_loc_reg(exprasmlist,op,location.size,right.location,location.register,mms_movescalar);
            location_release(exprasmlist,right.location);
          end;
      end;

end.
{
  $Log$
  Revision 1.5  2003-12-26 13:19:16  florian
    * rtl and compiler compile with -Cfsse2

  Revision 1.4  2003/12/26 00:32:22  florian
    + fpu<->mm register conversion

  Revision 1.3  2003/12/25 01:07:09  florian
    + $fputype directive support
    + single data type operations with sse unit
    * fixed more x86-64 stuff

  Revision 1.2  2003/12/23 14:38:07  florian
    + second_floataddsse implemented

  Revision 1.1  2003/10/13 01:58:04  florian
    * some ideas for mm support implemented
}