LI76642: INCORRECT DFP INSTRUCTION IN -QLIST

XL C/C++ for Linux Fix Pack 9 (August 2013 Update) for 11.1
June 2012 Update for XL C/C++ for Linux, V11.1
October 2012 Update for XL C/C++ for Linux, V11.1
XL C/C++ for Linux Fix Pack 8 (June 2013 Update) for 11.1

APAR status

• Closed as program error.

Error description

• The DFP instruction reported in the compiler listing
  (-qlist) output for the __d64_test_significance and
  __d128_test_significance compiler built-in functions are
  incorrect.  The instructions generated in the object file
  are correct, and DBX and the disassembler can recognize and
  decode the instructions correctly.  Only the compiler listing is
  reporting them as different DFP instructions.
  
  
        =====   EXAMPLE   =====
  
  TEST CASE FUNCTION:
  bool testSig(_Decimal128 dfp)
  {
     return __d128_test_significance(dfp, 31ULL) > 0;
  }
  
  
  
  LISTING OUTPUT:
   | 000000                    PDEF    testSig(_Decimal128)
   | 000000                    AKA      testSig__FDr
  0|                           PROC     dfp,fp2,fp3
  5| 000000 addi    3800001F 1 LI      gr0=31
  5| 000004 lfd     C821FFF8 1 LDFL    fp1=#MX_CONVF1_0(gr1,-8)
  5| 000008 std     F801FFF0 1 ST8     #MX_CONVF1_0(gr1,-16)=gr0
  5| 00000C lfd     C801FFF0 1 LDFL    fp0=#MX_CONVF1_0(gr1,-16)
  5| 000010 dtstexq FC001544 1 CDFSE   cr0=fp2,fp3,fp0,fp1
            ^^^^^^^<==== Incorrectly reported as 'dtstexq'
  5| 000014 bc      4180000C 1 BT      CL.4,cr0,0x1/lt,taken=50%
  5| 000018 addi    38600000 1 LI      gr3=0
  0| 00001C bclr    4E800020 1 BA      lr
  
  
  DBX OUTPUT:
  0x0 (testSig(_Decimal128))       3800001f          li   r0,0x1f
  0x4 (???)  c821fff8         lfd   fr1,-8(r1)
  0x8 (???)  f801fff0         std   r0,-16(r1)
  0xc (???)  c801fff0         lfd   fr0,-16(r1)
  0x10 (???) fc001544     dtstsfq   cr0,fr0,fr2
                          ^^^^^^^<=Correctly reported as 'dtstsfq'
  0x14 (???) 4180000c         blt   0x20 (???)
  0x18 (???) 38600000          li   r3,0x0
  0x1c (???) 4e800020         blr
  0x20 (???) 38600001          li   r3,0x1
  0x24 (???) 4e800020         blr
  

Local fix

• Use DBX or disassembler to verify instruction.
  

Problem summary

• PROBLEM DESCRIPTION:
  Incorrect instruction name in listing.
  
  USERS AFFECTED:
  Programs containing the decimal floating point
  __d64_test_significance and __d128_test_significance built in
  functions.
  

Problem conclusion

• Listing now shows correct instruction names.
  

Temporary fix

Comments

APAR Information

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

  IV11642

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

Fix information

• Fixed component name

  XL C/C++ FOR LI

• Fixed component ID

  5724X1400

Applicable component levels

• RB10 PSY

     UP