Conference caldec::wrl_atom

297.0. "Problems with debug info for V2.29 using 4.0A (Rev. 464)" by ROCK::EDWARDS () Tue Apr 22 1997 14:13

I am having some trouble with corruption of the debug information for line
numbers of procedures. If I use dis to get a disassembly of one of my analysis
routines I get:

dis -p EVSIM_queue_mem_instr $EV7P/evsimdbg7.anal.o | more
        EVSIM_queue_mem_instr:
  [sim/evsim.c: 3076] 0x100073cc:       27bb0008        ldah    gp, 8(t12)
  [sim/evsim.c: 3076] 0x100073d0:       23bdac64        lda     gp, -21404(gp)
  [sim/evsim.c: 3076] 0x100073d4:       23deff70        lda     sp, -144(sp)
  [sim/evsim.c: 3076] 0x100073d8:       b75e0008        stq     ra, 8(sp)
  [sim/evsim.c: 3076] 0x100073dc:       b53e0010        stq     s0, 16(sp)
  [sim/evsim.c: 3076] 0x100073e0:       b21e0078        stl     a0, 120(sp)
  [sim/evsim.c: 3076] 0x100073e4:       b63e0070        stq     a1, 112(sp)
  [sim/evsim.c: 3112] 0x100073e8:       a03e0078        ldl     t0, 120(sp)
  [sim/evsim.c: 3112] 0x100073ec:       4c211401        mulq    t0, 0x8, t0
  [sim/evsim.c: 3112] 0x100073f0:       a79d8030        ldq     at, -32720(gp)
  [sim/evsim.c: 3112] 0x100073f4:       43810401        addq    at, t0, t0

If I then call atom:

 atom ijpeg $EV7P/evsimdbg7.inst.o $EV7P/evsimdbg7.anal.o -o ijpeg.3pp -g

and then:

 dis -p EVSIM_queue_mem_instr ijpeg.3pp | more
ldreadst: WARNING iline index(96425) exceeds ilineMax(96425)
ldreadst: WARNING ilineMax was too small; It should have been 96426
        EVSIM_queue_mem_instr:
  [sim/evsim.c: 3083] 0x12001267c:      27bb1f0c        ldah    gp, 7948(t12)
  [sim/evsim.c: 3083] 0x120012680:      23bdfeb4        lda     gp, -332(gp)
  [sim/evsim.c: 3083] 0x120012684:      23deff70        lda     sp, -144(sp)
  [sim/evsim.c: 3083] 0x120012688:      b75e0008        stq     ra, 8(sp)
  [sim/evsim.c: 3083] 0x12001268c:      b53e0010        stq     s0, 16(sp)
  [sim/evsim.c: 3083] 0x120012690:      b21e0078        stl     a0, 120(sp)
  [sim/evsim.c: 3083] 0x120012694:      b63e0070        stq     a1, 112(sp)
  [sim/evsim.c: 3083] 0x120012698:      a03e0078        ldl     t0, 120(sp)
  [sim/evsim.c: 3083] 0x12001269c:      4c211401        mulq    t0, 0x8, t0
  [sim/evsim.c: 3083] 0x1200126a0:      a79d8030        ldq     at, -32720(gp)
  [sim/evsim.c: 3083] 0x1200126a4:      43810401        addq    at, t0, t0
  [sim/evsim.c: 3083] 0x1200126a8:      a421ffa0        ldq     t0, -96(t0)
  [sim/evsim.c: 3084] 0x1200126ac:      b43e0018        stq     t0, 24(sp)
  [sim/evsim.c: 3084] 0x1200126b0:      a21e0078        ldl     a0, 120(sp)
  [sim/evsim.c: 3084] 0x1200126b4:      4e009410        mulq    a0, 0x4, a0
  [sim/evsim.c: 3084] 0x1200126b8:      a79d8030        ldq     at, -32720(gp)
  [sim/evsim.c: 3084] 0x1200126bc:      43900410        addq    at, a0, a0

dis is not the only thing that gets confused, as the source level debuggers are
pretty useless when they display the wrong lines.

Of course I can supply all of these files....

Bruce

Bruce -

I am currently investigating that problem.  Another user also reported it with
hiprof and -g.  There is already a QAR open on the problem.  I will let you 
know when I have the fix in place.  

Gail Lyons

[Posted by WWW Notes gateway]