LI77362: USE LFIWAX/LFIWZX FOR INT TO FLOAT CONVERSION

April 2013 Update for XL C/C++ for Linux, V12.1
XL C/C++ for Linux Fix Pack 8 (October 2015 Update) for 12.1
XL C/C++ for Linux Fix Pack 4 (October 2013 Update) for 12.1
XL C/C++ for Linux Fix Pack 5 (December 2013 Update) for 12.1
XL C/C++ for Linux Fix Pack 6 (February 2014 Update) for 12.1
XL C/C++ for Linux Fix Pack 7 (May 2014 Update) for 12.1

APAR status

• Closed as program error.

Error description

• Compiler is not using lfiwax/lfiwzx assembler instruction
  for int to double conversion.
  
  Consider the following testcase:
  
  $cat intToDbl.C
  
  /******************************/
  
  double intToDbl(signed int *in)
  {
    return (double)(*in);
  }
  
  double intToDbl(signed int in)
  {
    return (double)(in);
  }
  
  double uintToDbl(unsigned int *in)
  {
    return (double)(*in);
  }
  
  double uintToDbl(unsigned int in)
  {
    return (double)(in);
  }
  /******************************/
  
  Compile the source file as:
  
  xlC -c -q64 -qlist -O2 intToDbl.C -qarch=pwr7
  
  
  Review output in intToDbl.lst file:
  
  ...
  ...
  
       | 000000                           PDEF     intToDbl(int *)
       | 000000                           AKA       intToDbl__FPi
      3|                                  PROC      in,gr3
      5| 000000 lwa      E8030002   1     L4A
  gr0=(int)(gr3,0)
      5| 000004 std      F801FFF0   1     ST8
  #MX_CONVF1_0(gr1,-16)=gr0
      5| 000008 lfd      C801FFF0   1     LFL
  vs0=#MX_CONVF1_0(gr1,-16)
      5| 00000C xscvsxdd F02005E0   1     FCFID     vs1=vs0,fcr
  
  
  Instead, the compiler should generate a lfiwax + xscvsxdd
  sequence:
  
  | 000000                           PDEF     intToDbl(int *)
       | 000000                           AKA       intToDbl__FPi
      1|                                  PROC      in,gr3
      3| 000000 lfiwax   7C001EAE   1     LFIWAX
  vs0=(int)(gr3,0)
      3| 000004 xscvsxdd F02005E0   1     FCFID     vs1=vs0,fcr
  

Local fix

• Not applicable
  

Problem summary

• compiler does not uses lfiwax/lfiwzx instructions for int to
  float conversion with -q64
  

Problem conclusion

• Relaxing the opportunity check code to use new instruction with
  -q64
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  IV37628

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  XL C/C++ FOR LI

• Fixed component ID

  5725C7300

Applicable component levels

• RC10 PSN IV37628

     UP06/09/13