Still more on the optimization ICE (as if anyone were interested)

[Bruce Korb <bkorb at sco dot COM>]

The other day, I merely compiled the stmt.i pre-processed
file.  I did not look at the object file:

$ elfinfo stmt.o
  section-name        size
  ================  ======
  .note                 28
  .symtab               32
  .strtab               23
  ================  ======
  Section Totals        83
  Total File Size      296


However, with the additional step of creating the assembly
output and *then* compiling:

$ stage1/xgcc -c -O2 stmt.s
$ elfinfo -S stmt.o
  section-name        size   Check Sum
  ================  ======  ==========
  .text              44234  0x37686DAF
  .data                 36  0x8C0A5C4F
  .rodata             2219  0xA711004C
  .bss                 360  0x00000000  (empty)
  .comment              40  0x62C3794B
  .note                 28  0x6C6FA773
  .rel.text          18048  0x6CED15DC
  .symtab             5296  0x125EC7E2
  .strtab             5297  0x142FD2FD
  ================  ======  ==========
  Section Totals     75558
  In-Use Totals      75198
  Total File Size    75704

the result looks more reasonable....it also links cc1 correctly.