Major debugging breakage

[law@redhat.com]

  > > Something has been hosed with line numbers for a long time;
  > 
  >     Just out of interest, would specifying the compiler flags
  >     ``-mpreferred-stack-boundary=2 -malign-functions=4'' increase the
  >     likelihood of hosing line-number tracking?
Not in the case that I'm looking at; the ones I'm trying to address
are strictly due to front-end bugs introduced after gcc-2.95.x.

  >     I've recently included these flags in our project at work and had
  >     one particular other programmer notice that the line tracking when
  >     debugging turned completely bogus.  Removing all optimisation flags
  >     and the aforementioned alignment flags seemed to rectify the problem
A number of optimizations can do "funny" things to source-line mappings,
but I'm not aware of cases where the preferred stack boundary or function
boundaries would affect the line table.

jeff