fpc/compiler/arm/narmcnv.pas

{
    Copyright (c) 1998-2002 by Florian Klaempfl

    Generate ARM assembler for type converting nodes

    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.

 ****************************************************************************
}
unit narmcnv;

{$i fpcdefs.inc}

interface

    uses
      node,ncnv,ncgcnv;

    type
       tarmtypeconvnode = class(tcgtypeconvnode)
         protected
           function first_int_to_real: tnode;override;
           function first_real_to_real: tnode;override;
           procedure second_int_to_real;override;
           procedure second_int_to_bool;override;
       end;

implementation

   uses
      verbose,globtype,globals,
      systems,
      symdef,aasmbase,aasmtai,aasmdata,symtable,
      defutil,
      cgbase,cgutils,
      pass_1,pass_2,procinfo,ncal,
      ncgutil,
      cpubase,cpuinfo,aasmcpu,cgobj,hlcgobj,cgcpu;


{*****************************************************************************
                             FirstTypeConv
*****************************************************************************}

    function tarmtypeconvnode.first_int_to_real: tnode;
      var
        fname: string[19];
      begin
        if (cs_fp_emulation in current_settings.moduleswitches) or
{$ifdef cpufpemu}
          (current_settings.fputype=fpu_soft) or
{$endif cpufpemu}
          (not(FPUARM_HAS_VFP_DOUBLE in fpu_capabilities[current_settings.fputype]) and
           not(FPUARM_HAS_FPA in fpu_capabilities[current_settings.fputype])) then
          result:=inherited first_int_to_real
        else
          begin
            { converting a 64bit integer to a float requires a helper }
            if is_64bitint(left.resultdef) or
              is_currency(left.resultdef) then
              begin
                { hack to avoid double division by 10000, as it's
                  already done by typecheckpass.resultdef_int_to_real }
                if is_currency(left.resultdef) then
                  left.resultdef := s64inttype;
                if is_signed(left.resultdef) then
                  fname := 'fpc_int64_to_double'
                else
                  fname := 'fpc_qword_to_double';
                result := ccallnode.createintern(fname,ccallparanode.create(
                  left,nil));
                left:=nil;
                if (tfloatdef(resultdef).floattype=s32real) then
                  inserttypeconv(result,s32floattype);
                firstpass(result);
                exit;
              end
            else
              { other integers are supposed to be 32 bit }
              begin
                if is_signed(left.resultdef) then
                  inserttypeconv(left,s32inttype)
                else
                  inserttypeconv(left,u32inttype);
                firstpass(left);
              end;
            result := nil;
            case current_settings.fputype of
              fpu_fpa,
              fpu_fpa10,
              fpu_fpa11:
                expectloc:=LOC_FPUREGISTER;
              else if FPUARM_HAS_VFP_EXTENSION in fpu_capabilities[current_settings.fputype] then
                expectloc:=LOC_MMREGISTER
              else
                internalerror(2009112702);
            end;
          end;
      end;


    function tarmtypeconvnode.first_real_to_real: tnode;
      begin
        if (current_settings.fputype=fpu_soft) and
           not (target_info.system in systems_wince) then
          begin
            case tfloatdef(left.resultdef).floattype of
              s32real:
                case tfloatdef(resultdef).floattype of
                  s64real:
                    result:=ctypeconvnode.create_explicit(ccallnode.createintern('float32_to_float64',ccallparanode.create(
                      ctypeconvnode.create_internal(left,search_system_type('FLOAT32REC').typedef),nil)),resultdef);
                  s32real:
                    begin
                      result:=left;
                      left:=nil;
                    end;
                  else
                    internalerror(2006101504);
                end;
              s64real:
                case tfloatdef(resultdef).floattype of
                  s32real:
                    result:=ctypeconvnode.create_explicit(ccallnode.createintern('float64_to_float32',ccallparanode.create(
                      ctypeconvnode.create_internal(left,search_system_type('FLOAT64').typedef),nil)),resultdef);
                  s64real:
                    begin
                      result:=left;
                      left:=nil;
                    end;
                  else
                    internalerror(2006101505);
                end;
              else
                internalerror(2006101506);
            end;
            left:=nil;
            firstpass(result);
            exit;
          end
        else
          Result := inherited first_real_to_real;
      end;


    procedure tarmtypeconvnode.second_int_to_real;
      const
        signedprec2vfppf: array[boolean,OS_F32..OS_F64] of toppostfix =
          ((PF_F32U32,PF_F64U32),
           (PF_F32S32,PF_F64S32));
      var
        instr : taicpu;
        href : treference;
        l1,l2 : tasmlabel;
        hregister : tregister;
        signed : boolean;
      begin
        case current_settings.fputype of
          fpu_fpa,
          fpu_fpa10,
          fpu_fpa11:
            begin
              { convert first to double to avoid precision loss }
              location_reset(location,LOC_FPUREGISTER,def_cgsize(resultdef));
              hlcg.location_force_reg(current_asmdata.CurrAsmList,left.location,left.resultdef,u32inttype,true);
              location.register:=cg.getfpuregister(current_asmdata.CurrAsmList,location.size);
              instr:=taicpu.op_reg_reg(A_FLT,location.register,left.location.register);
              if is_signed(left.resultdef) then
                begin
                  instr.oppostfix:=cgsize2fpuoppostfix[def_cgsize(resultdef)];
                  current_asmdata.CurrAsmList.concat(instr);
                end
              else
                begin
                  { flt does a signed load, fix this }
                  case tfloatdef(resultdef).floattype of
                    s32real,
                    s64real:
                      begin
                        { converting dword to s64real first and cut off at the end avoids precision loss }
                        instr.oppostfix:=PF_D;
                        current_asmdata.CurrAsmList.concat(instr);

                        current_asmdata.getglobaldatalabel(l1);
                        current_asmdata.getjumplabel(l2);
                        reference_reset_symbol(href,l1,0,const_align(8),[]);

                        cg.a_reg_alloc(current_asmdata.CurrAsmList,NR_DEFAULTFLAGS);
                        current_asmdata.CurrAsmList.concat(Taicpu.op_reg_const(A_CMP,left.location.register,0));
                        cg.a_jmp_flags(current_asmdata.CurrAsmList,F_GE,l2);
                        cg.a_reg_dealloc(current_asmdata.CurrAsmList,NR_DEFAULTFLAGS);

                        hregister:=cg.getfpuregister(current_asmdata.CurrAsmList,OS_F64);
                        new_section(current_asmdata.asmlists[al_typedconsts],sec_rodata_norel,l1.name,const_align(8));
                        current_asmdata.asmlists[al_typedconsts].concat(Tai_label.Create(l1));
                        { I got this constant from a test program (FK) }
                        current_asmdata.asmlists[al_typedconsts].concat(Tai_const.Create_32bit($41f00000));
                        current_asmdata.asmlists[al_typedconsts].concat(Tai_const.Create_32bit(0));

                        cg.a_loadfpu_ref_reg(current_asmdata.CurrAsmList,OS_F64,OS_F64,href,hregister);
                        current_asmdata.CurrAsmList.concat(setoppostfix(taicpu.op_reg_reg_reg(A_ADF,location.register,hregister,location.register),PF_D));
                        cg.a_label(current_asmdata.CurrAsmList,l2);

                        { cut off if we should convert to single }
                        if tfloatdef(resultdef).floattype=s32real then
                          begin
                            hregister:=location.register;
                            location.register:=cg.getfpuregister(current_asmdata.CurrAsmList,location.size);
                            cg.a_reg_alloc(current_asmdata.CurrAsmList,NR_DEFAULTFLAGS);
                            current_asmdata.CurrAsmList.concat(setoppostfix(taicpu.op_reg_reg(A_MVF,location.register,hregister),PF_S));
                          end;
                      end;
                    else
                      internalerror(2004100307);
                  end;
              end;
            end;
          else if FPUARM_HAS_VFP_DOUBLE in fpu_capabilities[current_settings.fputype] then
            begin
              location_reset(location,LOC_MMREGISTER,def_cgsize(resultdef));
              signed:=left.location.size=OS_S32;
              hlcg.location_force_mmregscalar(current_asmdata.CurrAsmList,left.location,left.resultdef,false);
              if (left.location.size<>OS_F32) then
                internalerror(2009112703);
              if left.location.size<>location.size then
                location.register:=cg.getmmregister(current_asmdata.CurrAsmList,location.size)
              else
                location.register:=left.location.register;
              current_asmdata.CurrAsmList.concat(setoppostfix(taicpu.op_reg_reg(A_VCVT,
                location.register,left.location.register),
                signedprec2vfppf[signed,location.size]));
            end
          else if FPUARM_HAS_VFP_EXTENSION in fpu_capabilities[current_settings.fputype] then
            begin
              location_reset(location,LOC_MMREGISTER,def_cgsize(resultdef));
              signed:=left.location.size=OS_S32;
              hlcg.location_force_mmregscalar(current_asmdata.CurrAsmList,left.location,left.resultdef,false);
              if (left.location.size<>OS_F32) then
                internalerror(2009112704);
              if left.location.size<>location.size then
                location.register:=cg.getmmregister(current_asmdata.CurrAsmList,location.size)
              else
                location.register:=left.location.register;
              if signed then
                current_asmdata.CurrAsmList.concat(setoppostfix(taicpu.op_reg_reg(A_VCVT,location.register,left.location.register), PF_F32S32))
              else
                current_asmdata.CurrAsmList.concat(setoppostfix(taicpu.op_reg_reg(A_VCVT,location.register,left.location.register), PF_F32U32));
            end
          else
            { should be handled in pass 1 }
            internalerror(2019050909);
        end;
      end;


    procedure tarmtypeconvnode.second_int_to_bool;
      var
        hreg1,
        hregister : tregister;
        href      : treference;
        resflags  : tresflags;
        hlabel    : tasmlabel;
        newsize   : tcgsize;
      begin
         secondpass(left);
         if codegenerror then
          exit;

         { Explicit typecasts from any ordinal type to a boolean type }
         { must not change the ordinal value                          }
         if (nf_explicit in flags) and
            not(left.location.loc in [LOC_FLAGS,LOC_JUMP]) then
           begin
              location_copy(location,left.location);
              newsize:=def_cgsize(resultdef);
              { change of size? change sign only if location is LOC_(C)REGISTER? Then we have to sign/zero-extend }
              if (tcgsize2size[newsize]<>tcgsize2size[left.location.size]) or
                 ((newsize<>left.location.size) and (location.loc in [LOC_REGISTER,LOC_CREGISTER])) then
                hlcg.location_force_reg(current_asmdata.CurrAsmList,location,left.resultdef,resultdef,true)
              else
                location.size:=newsize;
              exit;
           end;

         { Load left node into flag F_NE/F_E }
         resflags:=F_NE;

         if (left.location.loc in [LOC_SUBSETREG,LOC_CSUBSETREG,LOC_SUBSETREF,LOC_CSUBSETREF]) then
           hlcg.location_force_reg(current_asmdata.CurrAsmList,left.location,left.resultdef,left.resultdef,true);

         case left.location.loc of
            LOC_CREFERENCE,
            LOC_REFERENCE :
              begin
                if left.location.size in [OS_64,OS_S64] then
                 begin
                   hregister:=cg.getintregister(current_asmdata.CurrAsmList,OS_INT);
                   cg.a_load_ref_reg(current_asmdata.CurrAsmList,OS_32,OS_32,left.location.reference,hregister);
                   href:=left.location.reference;
                   inc(href.offset,4);
                   tbasecgarm(cg).cgsetflags:=true;
                   cg.a_op_ref_reg(current_asmdata.CurrAsmList,OP_OR,OS_32,href,hregister);
                   tbasecgarm(cg).cgsetflags:=false;
                 end
                else
                 begin
                   hlcg.location_force_reg(current_asmdata.CurrAsmList,left.location,left.resultdef,left.resultdef,true);
                   tbasecgarm(cg).cgsetflags:=true;
                   cg.a_op_reg_reg(current_asmdata.CurrAsmList,OP_OR,left.location.size,left.location.register,left.location.register);
                   tbasecgarm(cg).cgsetflags:=false;
                 end;
              end;
            LOC_FLAGS :
              begin
                resflags:=left.location.resflags;
              end;
            LOC_REGISTER,LOC_CREGISTER :
              begin
                if left.location.size in [OS_64,OS_S64] then
                 begin
                   hregister:=cg.getintregister(current_asmdata.CurrAsmList,OS_32);
                   cg.a_load_reg_reg(current_asmdata.CurrAsmList,OS_32,OS_32,left.location.register64.reglo,hregister);
                   tbasecgarm(cg).cgsetflags:=true;
                   cg.a_op_reg_reg(current_asmdata.CurrAsmList,OP_OR,OS_32,left.location.register64.reghi,hregister);
                   tbasecgarm(cg).cgsetflags:=false;
                 end
                else
                 begin
                   tbasecgarm(cg).cgsetflags:=true;
                   cg.a_op_reg_reg(current_asmdata.CurrAsmList,OP_OR,left.location.size,left.location.register,left.location.register);
                   tbasecgarm(cg).cgsetflags:=false;
                 end;
              end;
            LOC_JUMP :
              begin
                hregister:=cg.getintregister(current_asmdata.CurrAsmList,OS_INT);
                current_asmdata.getjumplabel(hlabel);
                cg.a_label(current_asmdata.CurrAsmList,left.location.truelabel);
                cg.a_load_const_reg(current_asmdata.CurrAsmList,OS_INT,1,hregister);
                cg.a_jmp_always(current_asmdata.CurrAsmList,hlabel);
                cg.a_label(current_asmdata.CurrAsmList,left.location.falselabel);
                cg.a_load_const_reg(current_asmdata.CurrAsmList,OS_INT,0,hregister);
                cg.a_label(current_asmdata.CurrAsmList,hlabel);
                tbasecgarm(cg).cgsetflags:=true;
                cg.a_op_reg_reg(current_asmdata.CurrAsmList,OP_OR,OS_INT,hregister,hregister);
                tbasecgarm(cg).cgsetflags:=false;
              end;
            else
              internalerror(2003113002);
         end;
         { load flags to register }
         location_reset(location,LOC_REGISTER,def_cgsize(resultdef));
         hreg1:=cg.getintregister(current_asmdata.CurrAsmList,location.size);
         cg.g_flags2reg(current_asmdata.CurrAsmList,location.size,resflags,hreg1);
         cg.a_reg_dealloc(current_asmdata.CurrAsmList,NR_DEFAULTFLAGS);
         if (is_cbool(resultdef)) then
           cg.a_op_reg_reg(current_asmdata.CurrAsmList,OP_NEG,location.size,hreg1,hreg1);

{$ifndef cpu64bitalu}
         if (location.size in [OS_64,OS_S64]) then
           begin
             location.register64.reglo:=hreg1;
             location.register64.reghi:=cg.getintregister(current_asmdata.CurrAsmList,OS_32);
             if (is_cbool(resultdef)) then
               { reglo is either 0 or -1 -> reghi has to become the same }
               cg.a_load_reg_reg(current_asmdata.CurrAsmList,OS_32,OS_32,location.register64.reglo,location.register64.reghi)
             else
               { unsigned }
               cg.a_load_const_reg(current_asmdata.CurrAsmList,OS_32,0,location.register64.reghi);
           end
         else
{$endif cpu64bitalu}
           location.register:=hreg1;
      end;


begin
  ctypeconvnode:=tarmtypeconvnode;
end.